#Call transcription software

[precallai]

Integrating AI Call Transcription into Your VoIP or CRM System

In today’s hyper-connected business environment, customer communication is one of the most valuable assets a company possesses. Every sales call, support ticket, or service request contains rich data that can improve business processes—if captured and analyzed properly. This is where AI call transcription becomes a game changer. By converting voice conversations into searchable, structured text, businesses can unlock powerful insights. The real value, however, comes when these capabilities are integrated directly into VoIP and CRM systems, streamlining operations and enhancing customer experiences.

Why AI Call Transcription Matters

AI call transcription leverages advanced technologies such as Automatic Speech Recognition (ASR) and Natural Language Processing (NLP) to convert real-time or recorded voice conversations into text. These transcripts can then be used for:

Compliance and auditing

Agent performance evaluation

Customer sentiment analysis

CRM data enrichment

Automated note-taking

Keyword tracking and lead scoring

Traditionally, analyzing calls was a manual and time-consuming task. AI makes this process scalable and real-time.

Key Components of AI Call Transcription Systems

Before diving into integration, it’s essential to understand the key components of an AI transcription pipeline:

Speech-to-Text Engine (ASR): Converts audio to raw text.

Speaker Diarization: Identifies and separates different speakers.

Timestamping: Tags text with time information for playback syncing.

Language Modeling: Uses NLP to enhance context, punctuation, and accuracy.

Post-processing Modules: Cleans up the transcript for readability.

APIs/SDKs: Interface for integration with external systems like CRMs or VoIP platforms.

Common Use Cases for VoIP + CRM + AI Transcription

The integration of AI transcription with VoIP and CRM platforms opens up a wide range of operational enhancements:

Sales teams: Automatically log conversations, extract deal-related data, and trigger follow-up tasks.

Customer support: Analyze tone, keywords, and escalation patterns for better agent training.

Compliance teams: Use searchable transcripts to verify adherence to legal and regulatory requirements.

Marketing teams: Mine conversation data for campaign insights, objections, and buying signals.

Step-by-Step: Integrating AI Call Transcription into VoIP Systems

Step 1: Capture the Audio Stream

Most modern VoIP systems like Twilio, RingCentral, Zoom Phone, or Aircall provide APIs or webhooks that allow you to:

Record calls in real time

Access audio streams post-call

Configure cloud storage for call files (MP3, WAV)

Ensure that you're adhering to legal and privacy regulations such as GDPR or HIPAA when capturing and storing call data.

Step 2: Choose an AI Transcription Provider

Several commercial and open-source options exist, including:

Google Speech-to-Text

AWS Transcribe

Microsoft Azure Speech

AssemblyAI

Deepgram

Whisper by OpenAI (open-source)

When selecting a provider, evaluate:

Language support

Real-time vs. batch processing capabilities

Accuracy in noisy environments

Speaker diarization support

API response latency

Security/compliance features

Step 3: Transcribe the Audio

Using the API of your chosen ASR provider, submit the call recording. Many platforms allow streaming input for real-time use cases, or you can upload an audio file for asynchronous transcription.

Here’s a basic flow using an API:

python

CopyEdit

import requests

response = requests.post(

    "https://api.transcriptionprovider.com/v1/transcribe",

    headers={"Authorization": "Bearer YOUR_API_KEY"},

    json={"audio_url": "https://storage.yourvoip.com/call123.wav"}

)

transcript = response.json()

The returned transcript typically includes speaker turns, timestamps, and a confidence score.

Step-by-Step: Integrating Transcription with CRM Systems

Once you’ve obtained the transcription, you can inject it into your CRM platform (e.g., Salesforce, HubSpot, Zoho, GoHighLevel) using their APIs.

Step 4: Map Transcripts to CRM Records

You’ll need to determine where and how transcripts should appear in your CRM:

Contact record timeline

Activity or task notes

Custom transcription field

Opportunity or deal notes

For example, in HubSpot:

python

CopyEdit

requests.post(

    "https://api.hubapi.com/engagements/v1/engagements",

    headers={"Authorization": "Bearer YOUR_HUBSPOT_TOKEN"},

    json={

        "engagement": {"active": True, "type": "NOTE"},

        "associations": {"contactIds": [contact_id]},

        "metadata": {"body": transcript_text}

    }

)

Step 5: Automate Trigger-Based Actions

You can automate workflows based on keywords or intent in the transcript, such as:

Create follow-up tasks if "schedule demo" is mentioned

Alert a manager if "cancel account" is detected

Move deal stage if certain intent phrases are spoken

This is where NLP tagging or intent classification models can add value.

Advanced Features and Enhancements

1. Sentiment Analysis

Apply sentiment models to gauge caller mood and flag negative experiences for review.

2. Custom Vocabulary

Teach the transcription engine brand-specific terms, product names, or industry jargon for better accuracy.

3. Voice Biometrics

Authenticate speakers based on voiceprints for added security.

4. Real-Time Transcription

Show live captions during calls or video meetings for accessibility and note-taking.

Challenges to Consider

Privacy & Consent: Ensure callers are aware that calls are recorded and transcribed.

Data Storage: Securely store transcripts, especially when handling sensitive data.

Accuracy Limitations: Background noise, accents, or low-quality audio can degrade results.

System Compatibility: Some CRMs may require custom middleware or third-party plugins for integration.

Tools That Make It Easy

Zapier/Integromat: For non-developers to connect transcription services with CRMs.

Webhooks: Trigger events based on call status or new transcriptions.

CRM Plugins: Some platforms offer native transcription integrations.

Final Thoughts

Integrating AI call transcription into your VoIP and CRM systems can significantly boost your team’s productivity, improve customer relationships, and offer new layers of business intelligence. As the technology matures and becomes more accessible, now is the right time to embrace it.

With the right strategy and tools in place, what used to be fleeting conversations can now become a core part of your data-driven decision-making process.

[talesfromthornvale]

Tales from Thornvale Transcripts

hello couriers! my transcription project is here!

If you would like to read and/or print the transcripts for Season One of Tales from Thornvale you can now do so here:

(there is both a document you can read/download from google drive and signatures ready to print if you would like to bind it into a copy of your own)

full disclosure I have not transcribed before, there is a fair chance there's mistakes and/or errors, but this is labour of love that I hope you enjoy. (if you find mistakes feel free to let me know and I'll correct them)

[dduane]

And relatedly, how the HELL are you handling the seemingly inevitable wrist and arm issues that come with the craft? Last time I speed-wrote anything, my physical therapist called me a dozen kinds of idiot. Silently, because he’s a nice guy, but the expressions were an experience…

(I hope I do not overstep here: I’ve currently got a novel burning half my brain and a deeply unhappy right wrist, so it comes from a very proximate inspiration!)

The quick answer: because I too ran into the RSI barrier at an early stage, I've been using speech-to-text tech whenever possible, from when it very first became available in the early 1990s. @petermorwood and I were very early adopters of the first truly reliable STT software, Dragon Dictate (which eventually became Dragon Naturally Speaking).

We got Terry Pratchett hooked up to it, too, and thereby, if indirectly, made it possible for him to finish a book or two (or three...) more than would otherwise have been feasible for him. (Terry's own experts came in and fine-tuned the basic software for T's own needs.)

I've currently got the professional version of Dragon Naturally Speaking installed on my desktop machine, and can sit in my Comfy Chair with my feet up and dictate, watching the words spill out onto the big TV screen in the living room. Or alternately: due to currently being on the road a lot, I'm mostly using the app version of the software, Dragon Anywhere. Dictate to it, when hooked up to broadband, and it types what you're saying as you watch, with 95%-or-better accuracy out of the box. (And the program is endlessly configurable to handle specialized vocabulary, weird alien character names, or whatever.) When you're done, you save the file and the app'll email you a .doc-file transcription of what you just dictated. Cut and paste this into your preferred writing software and—having exuded a chunk of "zero draft" without excessive amounts of wear and tear on your sinews—you can then edit at your leisure.

The app runs on a relatively low-cost monthly subscription model... which makes it handily accessible to folks who can't afford the (unquestionably hefty) price tag on the standalone big-machine install of the full program. I recommend the app highly. You might consider trying it for a month or so and seeing how it works for you. The subscription goes month-to-month, and is easily cancelled if you don't care for it.

Anyway: hope this helps!

[xinganhao]

cherry on top 🍒 mafia boss!seungcheol x reader. (4)

stories like this always end with a damsel in distress. except—this time around—you’re not the one who needs saving. previous chapter + masterlist.

📄 Minutes of strategic information meeting, filed by Kim Mingyu (Mafia Soldier, Logistics & Recon)

Date: ██████████ Location: Safehouse Omega-9, Undisclosed City Perimeter Time: 03:17 HRS

ATTENDEES:

Yoon Jeonghan (Underboss)

Lee Chan (Combat Unit Leader)

Chwe Hansol (Surveillance Division)

Kim Mingyu (Logistics & Recon; Recording Officer)

Civilian Target [REDACTED] (Unauthorized Attendee)

AGENDA:

Contingency Plan for Retrieval of Boss (S.Coups)

Chain of Command During Absence

External Threat Assessment

[BEGIN TRANSCRIPT]

JEONGHAN: We go in through the east dock. Two snipers posted by 03:40. Chan leads breach. Hansol, your eyes stay on thermal—no improvisation this time.

HANSOL: I never improvise. My brilliance is structured.

CHAN: Can we not do this right now?

JEONGHAN: [ignoring them] Mingyu, once we get him out, you're on evac. Full blackout route. No trackers, no chatter.

MINGYU: Copy.

HANSOL: Any updates on who turned? Someone had to leak coordinates.

CHAN: There’s a list. We’ll handle it after we bring the boss home. One fire at a time.

[DOOR SLAMS OPEN. SOUND OF HIGH-HEELED FOOTSTEPS. SILENCE.]

CIVILIAN TARGET: You’re planning this without me?

JEONGHAN: [visibly tense] You weren’t invited.

CIVILIAN TARGET: He’s my belo—my boyfriend, Jeonghan. You think I’m just going to sit around while you play war games?

JEONGHAN: This isn’t a movie. You’re a civilian. You don’t belong in this room.

CIVILIAN TARGET: No, I’m the reason he still believes in soft things. I belong more than half the people at this table.

CHAN: She’s got a point.

JEONGHAN: Chan.

CHAN: I’m just saying. She’s not exactly fragile.

HANSOL: She did rewire one of my bugs with a paperclip. That was... not unimpressive.

JEONGHAN: [sighs] This isn’t about guts. It’s about blood.

CIVILIAN TARGET: Then you should know mine’s already on the line. Every second he’s gone, I feel it. And I’m done being sidelined. I’m not here to ask. I’m here to help.

[BEAT OF SILENCE. THEN—]

JEONGHAN: You get one job. And if you screw it up, I’ll personally drag you out.

CIVILIAN TARGET: Deal.

JEONGHAN: Hansol, give her the map. Mingyu, loop her in.

MINGYU: You’re going to need a comm. And a bulletproof vest.

CIVILIAN TARGET: Got both. And a knife in my boot.

CHAN: Okay, badass.

[MEETING CONTINUED UNDER LEVEL-2 SECRECY PROTOCOLS. TRANSCRIPT REDACTED. END OF MINUTES.]

FINAL NOTES:

Civilian Target formally added to Operation Homecoming roster.

Jeonghan authorized conditional field involvement.

Morale status: heightened.

Risk level: astronomically high.

🗂️ Operation Homecoming: Field Notes & Briefing Report, compiled by mafia underboss, Yoon Jeonghan

Clearance Level: Top Confidential Date Logged: ██████████ Location: Safehouse Omega-9

SUMMARY: Boss (S.Coups) was captured 48 hours ago following the receipt of a falsified emergency ping traced back to the civilian target’s encoded channel. The ping claimed she’d been injured and was en route to an undisclosed hospital in Sector D. According to surveillance logs, the Boss diverted course alone, abandoning standard security protocol. We believe he was intentionally isolated through signal jamming, then intercepted at the underpass beneath Route 14.

AUTOPSY OF THE TRAP:

Fake GPS tag mimicked civilian target’s bio-signal pattern

Voice distortion software replicated her distress call

EMP deployed upon vehicle arrival to disable tracking

Tactical unit waited with sedation-grade rounds

CURRENT LOCATION OF BOSS: Confirmed. Underground storage facility, formerly Syndicate-aligned. Defected cell now controls the zone. Reinforcements on site. Boss presumed alive—last thermal footage confirms faint movement.

INTERVENTION STRATEGY: OPERATION HOMECOMING

Phase One – Extraction:

Entry through east dock (03:40 HRS)

Chan leads breach unit, Hansol on thermal, Mingyu handling evac

All units silent channel only

Phase Two – Internal Sweep:

Civilian target assigned distraction and misdirection role (see below)

Two-minute window to locate and stabilize Boss

Phase Three – Extraction + Fade:

Mingyu initiates blackout route

Decoys deployed on west perimeter to delay pursuit

Rendezvous at Site Echo

CIVILIAN TARGET: PERFORMANCE LOG

Arrived wearing borrowed Kevlar and jeans tucked into combat boots. Asked if bulletproof vests same in women’s sizes. Did not wait for response.

Showed immediate enthusiasm, zero tactical finesse. Hansol gave her the map. She held it upside down. Twice.

Informed her she’d be working as the visual diversion. Her response: “Like bait?” Followed by: “Cool. I’m good at being annoying.”

Surprisingly effective. Created a loud enough ruckus on the perimeter to draw three guards off their posts. Managed to bluff her way past checkpoint by pretending to be a lost food delivery driver. Claimed she had gluten-free soba for a man named Kevin. There is no Kevin.

Still not sure how she pulled it off.

When Boss was found, he was semi-conscious but breathing. Whispered her name first.

END STATUS:

Boss retrieved.

Minimal casualties (1 injured – not fatal)

Facility compromised but not traced

Civilian target cried in the van. Then threatened to punch me for writing that down. I'm writing it down anyway.

FOOTNOTE — for Seungcheol’s eyes only: You’re reckless, stubborn, and impossible to reason with. But apparently, that’s your thing. You’re also luckier than most of us ever will be.

She didn’t sleep. Not once. Kept looking at every door like you might walk through it.

When you did, she didn’t even say anything. Just threw her arms around you like gravity stopped working.

Try not to make her go through that again.

– YJH

📱 Phone history log, filed by mafia soldier Chwe Hansol

Device: S.Coups' Personal Line (Encrypted Channel #017) Status: Outgoing Messages Only – Blocked by Signal Jammer Timestamp Range: ██:██–██:██ (Time of Abduction)

NOTE: Texts never reached intended recipient. Recovered during post-mission diagnostics. For archival purposes.

[01:12 AM] Where are you? They said you were hurt. I'm on my way.

[01:15 AM] Which hospital? No one's answering. This isn't funny. Call me.

[01:17 AM] Your signal keeps bouncing. Something's wrong. Stay where you are.

[01:21 AM] I swear to god if they laid a hand on you

[01:24 AM] No ambulance ever came.

[01:25 AM] This is a setup.

[01:27 AM] I'm so stupid. They used you. Fuck fuck fuck

[01:28 AM] I should've followed protocol. Should’ve sent Mingyu. Should’ve sent anyone but me.

[01:30 AM] If you get this, lock all the windows. Call Jeonghan. Stay put.

[01:34 AM] They knew I’d come for you.

[01:36 AM] This isn’t your fault.

[01:39 AM] Don’t come after me.

[01:41 AM] Love, beloved, please. Don’t try to save me.

[01:45 AM] You always do this—you throw yourself into fires you don't understand.

[01:49 AM] If they hurt you because of me, I’ll never forgive myself.

[01:52 AM] Tell Jeonghan to burn everything. Get out. Go far.

[01:54 AM] Forget me if you have to. Just live.

[02:01 AM] I love you. Please, please, please, don’t be stupid.

[END OF RECOVERED LOG]

📰 Excerpt from "The Ethics of Mafias: Love in the Line of Fire", a follow-up think piece by Xu Minghao

... If leadership within organized crime is already an ethical minefield, then love within it is something more volatile still: a paradox of vulnerability embedded in violence. New whispers surround the figure known only as S.Coups—the alleged mafia boss whose name, until recently, conjured images of discipline, domination, and an empire forged in precision.

Now, another narrative has emerged. One that reshapes how we understand not just the man, but the very myth he embodies.

According to rumors sourced from both within and outside the organization, S.Coups may have a romantic partner. Not a fellow operative, nor a political alliance. But a civilian. Someone unaffiliated and—crucially—untouched by the bloodied logic of the underworld.

If this is true, the implications are vast.

To love in his position is a risk. It is weakness, some would say. Yet others might argue that such love is the only thing capable of keeping a man like him from becoming monstrous. If the rumors are accurate, she is the reason he looks over his shoulder less. The reason he checks his own wrath. The reason his most trusted lieutenants have stopped fearing him and started worrying about him.

Love, here, is not a diversion. It is discipline.

And perhaps that is the most fascinating ethical twist of all: that this boss, so often theorized as either tyrant or savior, might be both—because of her.

Some say he texts her between assassinations. That he buys her gummy bears because she mentioned liking them once, months ago. That he has started folding her laundry and learning her aunt’s dietary restrictions. These are, of course, unconfirmed. They seem almost laughably mundane. But within the shadowed world of syndicates and secret wars, what could be more radical than tenderness?

Others claim that he was taken. There are now verified reports of a failed abduction and his eventual rescue. She was allegedly involved. They say she showed up unarmed, untrained, and utterly unafraid. They say she demanded to be part of the rescue mission. They say she was reckless, infuriating, and ultimately, instrumental.

And that when he saw her again, he wept.

To be loved, it turns out, is not always soft. Sometimes, it is brutal and inelegant and wildly inconvenient. But in the context of a life built on violence, to be loved is to be saved. Again and again. In the ways that matter.

Whether S.Coups is worthy of that love is not the question. The question is whether it has already changed him. Whether, in the end, the girl outside the syndicate might be the only thing real in a world made of smoke and mirrors.

And whether that, more than power or fear, will be his lasting legacy.

Mafia boss S.Coups is many things. Protector, manipulator. Brother, enemy, friend.

It seems we must add two more things:

Lover, and loved.

FIN. THANK YOU FOR READING CHERRY ON TOP!

› scroll through all my work ദ്ദി ˉ͈̀꒳ˉ͈́ )✧ ᶻ 𝗓 𐰁 .ᐟ my masterlist | @xinganhao

[mostlysignssomeportents]

Conspiratorialism as a material phenomenon

I'll be in TUCSON, AZ from November 8-10: I'm the GUEST OF HONOR at the TUSCON SCIENCE FICTION CONVENTION.

I think it behooves us to be a little skeptical of stories about AI driving people to believe wrong things and commit ugly actions. Not that I like the AI slop that is filling up our social media, but when we look at the ways that AI is harming us, slop is pretty low on the list.

The real AI harms come from the actual things that AI companies sell AI to do. There's the AI gun-detector gadgets that the credulous Mayor Eric Adams put in NYC subways, which led to 2,749 invasive searches and turned up zero guns:

https://www.cbsnews.com/newyork/news/nycs-subway-weapons-detector-pilot-program-ends/

Any time AI is used to predict crime – predictive policing, bail determinations, Child Protective Services red flags – they magnify the biases already present in these systems, and, even worse, they give this bias the veneer of scientific neutrality. This process is called "empiricism-washing," and you know you're experiencing it when you hear some variation on "it's just math, math can't be racist":

https://pluralistic.net/2020/06/23/cryptocidal-maniacs/#phrenology

When AI is used to replace customer service representatives, it systematically defrauds customers, while providing an "accountability sink" that allows the company to disclaim responsibility for the thefts:

https://pluralistic.net/2024/04/23/maximal-plausibility/#reverse-centaurs

When AI is used to perform high-velocity "decision support" that is supposed to inform a "human in the loop," it quickly overwhelms its human overseer, who takes on the role of "moral crumple zone," pressing the "OK" button as fast as they can. This is bad enough when the sacrificial victim is a human overseeing, say, proctoring software that accuses remote students of cheating on their tests:

https://pluralistic.net/2022/02/16/unauthorized-paper/#cheating-anticheat

But it's potentially lethal when the AI is a transcription engine that doctors have to use to feed notes to a data-hungry electronic health record system that is optimized to commit health insurance fraud by seeking out pretenses to "upcode" a patient's treatment. Those AIs are prone to inventing things the doctor never said, inserting them into the record that the doctor is supposed to review, but remember, the only reason the AI is there at all is that the doctor is being asked to do so much paperwork that they don't have time to treat their patients:

https://apnews.com/article/ai-artificial-intelligence-health-business-90020cdf5fa16c79ca2e5b6c4c9bbb14

My point is that "worrying about AI" is a zero-sum game. When we train our fire on the stuff that isn't important to the AI stock swindlers' business-plans (like creating AI slop), we should remember that the AI companies could halt all of that activity and not lose a dime in revenue. By contrast, when we focus on AI applications that do the most direct harm – policing, health, security, customer service – we also focus on the AI applications that make the most money and drive the most investment.

AI hasn't attracted hundreds of billions in investment capital because investors love AI slop. All the money pouring into the system – from investors, from customers, from easily gulled big-city mayors – is chasing things that AI is objectively very bad at and those things also cause much more harm than AI slop. If you want to be a good AI critic, you should devote the majority of your focus to these applications. Sure, they're not as visually arresting, but discrediting them is financially arresting, and that's what really matters.

All that said: AI slop is real, there is a lot of it, and just because it doesn't warrant priority over the stuff AI companies actually sell, it still has cultural significance and is worth considering.

AI slop has turned Facebook into an anaerobic lagoon of botshit, just the laziest, grossest engagement bait, much of it the product of rise-and-grind spammers who avidly consume get rich quick "courses" and then churn out a torrent of "shrimp Jesus" and fake chainsaw sculptures:

https://www.404media.co/email/1cdf7620-2e2f-4450-9cd9-e041f4f0c27f/

For poor engagement farmers in the global south chasing the fractional pennies that Facebook shells out for successful clickbait, the actual content of the slop is beside the point. These spammers aren't necessarily tuned into the psyche of the wealthy-world Facebook users who represent Meta's top monetization subjects. They're just trying everything and doubling down on anything that moves the needle, A/B splitting their way into weird, hyper-optimized, grotesque crap:

https://www.404media.co/facebook-is-being-overrun-with-stolen-ai-generated-images-that-people-think-are-real/

In other words, Facebook's AI spammers are laying out a banquet of arbitrary possibilities, like the letters on a Ouija board, and the Facebook users' clicks and engagement are a collective ideomotor response, moving the algorithm's planchette to the options that tug hardest at our collective delights (or, more often, disgusts).

So, rather than thinking of AI spammers as creating the ideological and aesthetic trends that drive millions of confused Facebook users into condemning, praising, and arguing about surreal botshit, it's more true to say that spammers are discovering these trends within their subjects' collective yearnings and terrors, and then refining them by exploring endlessly ramified variations in search of unsuspected niches.

(If you know anything about AI, this may remind you of something: a Generative Adversarial Network, in which one bot creates variations on a theme, and another bot ranks how closely the variations approach some ideal. In this case, the spammers are the generators and the Facebook users they evince reactions from are the discriminators)

https://en.wikipedia.org/wiki/Generative_adversarial_network

I got to thinking about this today while reading User Mag, Taylor Lorenz's superb newsletter, and her reporting on a new AI slop trend, "My neighbor’s ridiculous reason for egging my car":

https://www.usermag.co/p/my-neighbors-ridiculous-reason-for

The "egging my car" slop consists of endless variations on a story in which the poster (generally a figure of sympathy, canonically a single mother of newborn twins) complains that her awful neighbor threw dozens of eggs at her car to punish her for parking in a way that blocked his elaborate Hallowe'en display. The text is accompanied by an AI-generated image showing a modest family car that has been absolutely plastered with broken eggs, dozens upon dozens of them.

According to Lorenz, variations on this slop are topping very large Facebook discussion forums totalling millions of users, like "Movie Character…,USA Story, Volleyball Women, Top Trends, Love Style, and God Bless." These posts link to SEO sites laden with programmatic advertising.

The funnel goes:

i. Create outrage and hence broad reach;

ii, A small percentage of those who see the post will click through to the SEO site;

iii. A small fraction of those users will click a low-quality ad;

iv. The ad will pay homeopathic sub-pennies to the spammer.

The revenue per user on this kind of scam is next to nothing, so it only works if it can get very broad reach, which is why the spam is so designed for engagement maximization. The more discussion a post generates, the more users Facebook recommends it to.

These are very effective engagement bait. Almost all AI slop gets some free engagement in the form of arguments between users who don't know they're commenting an AI scam and people hectoring them for falling for the scam. This is like the free square in the middle of a bingo card.

Beyond that, there's multivalent outrage: some users are furious about food wastage; others about the poor, victimized "mother" (some users are furious about both). Not only do users get to voice their fury at both of these imaginary sins, they can also argue with one another about whether, say, food wastage even matters when compared to the petty-minded aggression of the "perpetrator." These discussions also offer lots of opportunity for violent fantasies about the bad guy getting a comeuppance, offers to travel to the imaginary AI-generated suburb to dole out a beating, etc. All in all, the spammers behind this tedious fiction have really figured out how to rope in all kinds of users' attention.

Of course, the spammers don't get much from this. There isn't such a thing as an "attention economy." You can't use attention as a unit of account, a medium of exchange or a store of value. Attention – like everything else that you can't build an economy upon, such as cryptocurrency – must be converted to money before it has economic significance. Hence that tooth-achingly trite high-tech neologism, "monetization."

The monetization of attention is very poor, but AI is heavily subsidized or even free (for now), so the largest venture capital and private equity funds in the world are spending billions in public pension money and rich peoples' savings into CO2 plumes, GPUs, and botshit so that a bunch of hustle-culture weirdos in the Pacific Rim can make a few dollars by tricking people into clicking through engagement bait slop – twice.

The slop isn't the point of this, but the slop does have the useful function of making the collective ideomotor response visible and thus providing a peek into our hopes and fears. What does the "egging my car" slop say about the things that we're thinking about?

Lorenz cites Jamie Cohen, a media scholar at CUNY Queens, who points out that subtext of this slop is "fear and distrust in people about their neighbors." Cohen predicts that "the next trend, is going to be stranger and more violent.”

This feels right to me. The corollary of mistrusting your neighbors, of course, is trusting only yourself and your family. Or, as Margaret Thatcher liked to say, "There is no such thing as society. There are individual men and women and there are families."

We are living in the tail end of a 40 year experiment in structuring our world as though "there is no such thing as society." We've gutted our welfare net, shut down or privatized public services, all but abolished solidaristic institutions like unions.

This isn't mere aesthetics: an atomized society is far more hospitable to extreme wealth inequality than one in which we are all in it together. When your power comes from being a "wise consumer" who "votes with your wallet," then all you can do about the climate emergency is buy a different kind of car – you can't build the public transit system that will make cars obsolete.

When you "vote with your wallet" all you can do about animal cruelty and habitat loss is eat less meat. When you "vote with your wallet" all you can do about high drug prices is "shop around for a bargain." When you vote with your wallet, all you can do when your bank forecloses on your home is "choose your next lender more carefully."

Most importantly, when you vote with your wallet, you cast a ballot in an election that the people with the thickest wallets always win. No wonder those people have spent so long teaching us that we can't trust our neighbors, that there is no such thing as society, that we can't have nice things. That there is no alternative.

The commercial surveillance industry really wants you to believe that they're good at convincing people of things, because that's a good way to sell advertising. But claims of mind-control are pretty goddamned improbable – everyone who ever claimed to have managed the trick was lying, from Rasputin to MK-ULTRA:

https://pluralistic.net/HowToDestroySurveillanceCapitalism

Rather than seeing these platforms as convincing people of things, we should understand them as discovering and reinforcing the ideology that people have been driven to by material conditions. Platforms like Facebook show us to one another, let us form groups that can imperfectly fill in for the solidarity we're desperate for after 40 years of "no such thing as society."

The most interesting thing about "egging my car" slop is that it reveals that so many of us are convinced of two contradictory things: first, that everyone else is a monster who will turn on you for the pettiest of reasons; and second, that we're all the kind of people who would stick up for the victims of those monsters.

Tor Books as just published two new, free LITTLE BROTHER stories: VIGILANT, about creepy surveillance in distance education; and SPILL, about oil pipelines and indigenous landback.

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2024/10/29/hobbesian-slop/#cui-bono

Image: Cryteria (modified) https://commons.wikimedia.org/wiki/File:HAL9000.svg

CC BY 3.0 https://creativecommons.org/licenses/by/3.0/deed.en

[transmutationisms]

Not sure if I’m understanding what a medical scribe does but do you have to lie for them? lol Bc I always look at the notes from visits and they will claim they discussed like 20 different risks, factors etc that they straight up never mentioned, for liability reasons I assume, but also the general inaccuracy is staggering even if they aren’t mischaracterizing me to be an asshole. Providers I consider thoughtful and trustworthy despite their position do this as well, truly everyone. Or is it a research setting you’re in? I am unphased by this kind of thing but I’m also curious ab your insider take on what is happening exactly/ free rant prompt?

the physicians i've worked with honestly barely even check what their scribes write & i routinely put notes in assessment/plan that are more honest than they would prefer (for example, i paste in guild recommendations on vaccination when i scribe for antivaxx physicians, because i think patients should have actual accurate information if and when they check their own visit records).

these days, some practitioners do their own scribing, some use human scribes (this is typically a person in black scrubs sitting in the corner of your appointment with a laptop who will be introduced as such), and some use 'ai' transcription software. the software solution is cheaper but has a tendency to transcribe certain things the physician doesn't want documented, like when they have a non-medical chat with the patient about the big game. i don't know how well the software handles other issues that require a judgment call, honestly. in any case, if you're already reading your own chart records, you can determine whether a scribe is writing your notes by scrolling down to the bottom because scribes are required (i believe this is basically universal worldwide lol) to sign/attest to our notes, so that person's name or the name of the software will be there above the physician's e-signature. but final signature of chart notes is always done by a practitioner (either the physician or a nurse-practitioner if that's who you're seeing), and that person has final authority to edit the notes if they want to.

notes on risk factors and general health counselling are i believe usually not a main factor relevant to malpractice determinations (they're written more to communicate to future practitioners, though this is kind of a joke because doctors rarely spend long even reading patients' charts, esp in routine/outpatient care). this obviously varies though depending on what jurisdiction you're in and how your healthcare system is funded, and the question of how tort reform and no-fault compensation affect both the practice and notation of 'defensive medicine' is a whole thing lol. but i might be able to give some insight into other reasons why this mismatch can occur:

-physicians give really vague and cursory advice and scribe it as though it was a whole conversation. eg, a doctor might glance at your chart and say "well, you're young and healthy so i'm not worried about heart disease" and then note this in the a&p as "Patient was counselled on risk factors for CAD and should continue to eat a balanced diet, avoid smoking, and exercise regularly. Patient will follow up in one year or sooner as needed" and the patient is meanwhile walking away only thinking "i guess my heart is fine" or "that exam sure was uncomfortable"

-in many healthcare systems, you may be asked to fill out certain screening questionnaires (eg for depression, anxiety, life-style risk factors, smoking habits, &c &c) and these may be getting scribed as "Advised on [risk factor or health behaviour]" even if all the doctor did was glance at it and say Okay. or not even do that

-similarly, if you got handed or offered pamphlets, this might be scribed as an entire conversation lol. like it shouldn't be. but it does happen.

-in electronic charts, routine notes like advice on risk factors are typically not even typed out by hand anymore, they have what are usually called 'dot-phrases' and may even be auto-entered before a visit as part of the chart template ('dot-notes'). for a number of reasons, dot-phrases can get entered for routine concerns even if they didn't actually come up with a specific patient.

-for example, doctors who take their own notes may enter that they advised on xyz because they typically do, and intended to, but literally just forgot. scribes provide another layer of checking to avoid this type of thing but plenty of lazy or incompetent scribes may also do this. this is exacerbated by poor clinic flow that shrinks appointment times and double- or triple-books appointments (which is in turn exacerbated by underfunded national health systems, and private equity ownership of private clinics)

-or, and again this depends where you are, but some national healthcare systems or insurance carriers actually require certain counselling to be given across the board, and your doctor may literally just be documenting it by law even if they think it's stupid or harmful or irrelevant to you in particular and didn't discuss it on purpose. for example, this happened with the last doctor i scribed for because he disagreed with the government requirements about documenting and categorising bmi, but in order to get the patients reimbursed for the visit, the charts simply had to include the dot-phrases calculating bmi and calling the patients 'out of range' if they were over 23 or 25.

[hb-writes]

Voicemail

Day 6 prompt (600 words) for 10 days to 1K: Voicemail

Fandom: The Pitt

Characters: Dr. Robby-mentioned, Maya Bennett (OC)

—

Maya had been drifting off when her phone started buzzing, the vibration against the desk startling her and garnering the attention of everyone in Professor Church’s lecture. She hadn’t been paying the slightest bit of attention, but all eyes were on her as she scrambled to send the call to voicemail, her heart rate spiking as she noted the name on the screen. 

“Sorry!” Maya offered, heat blooming in her cheeks as her professor resumed his lecture. Her classmates’s attention followed, allowing Maya to focus on the voice to text transcription of Robby’s voicemail. By the end of the message, a heavy rock of dread had dropped in Maya’s stomach. 

She thought that odd considering she’d spent half the day filled with dread. For a few hours now, Maya had expected a call from the man she thought of as a father, ever since she’d accidentally answered Jake’s facetime, giving him an unintended view of the bruises she’d spent the last four days hiding. 

Maya had been so careful—avoiding Robby or Jake or anyone who worked with her dad in the Pitt since Sunday. She had presented herself as exceedingly busy between work and school, easily explaining away the uncharacteristic delays in answering texts and calls. 

But she hadn’t been busy with work or school—in fact, Maya hadn’t left the house in days. She’d spent most of her time sleeping and what little time she spent awake, she’d used to manage her absences and keep up appearances. Looking at her phone or laptop for too long made her headache and nausea worse, so mostly, Maya had spent the last few days napping and just getting by. 

Thankfully, the nausea had dissipated this morning, only for her call with Jake to make her want to throw up all over again. For the first time in ages, bribery hadn’t been enough to sway her pseudo-brother from tattling. 

Maya was actually surprised it had taken this long for Robby’s call to come—she’d talked with Jake just before noon when her roommate had come home on lunch break to help make Maya presentable, covering the bruising with a heavy layer of expertly applied makeup. Either Jake had waited a bit to tell Robby or her dad was having a hell of a day and he was just getting to her now.

She hoped it was the first one. 

Robby didn’t usually order Maya around, especially now that she was an adult. She was in grad school. She worked. She’d moved out. She paid most of her own bills, and though she sometimes only took care of “adulty” stuff because Robby gave her firm-ish reminders, she generally managed things on her own. 

But as Maya reviewed the transcription again, her imagination delivering the words in Robby’s voice, complete with the assumed tone and pauses and all the little sounds the transcription software didn’t catch, she determined it was very clearly an order and he was very clearly unhappy. 

More dread settled in her then which seemed impossible considering how much she already carried. Maya had been considering skipping her last class and heading home for the day, but the voicemail gave her pause. Maya knew she’d have to face Robby eventually, but she needed time to collect herself first. And she wanted to call Jake—to yell at him and to do recon. Even though she could imagine how mad Robby was in general, it was better to know how Jake’s conversation had gone. She needed to know a little more than Robby’s message:

I spoke with Jake. Come to the Pitt when you get this.

—

The Pitt Masterlist

Part 2

[mariacallous]

On Saturday, an Associated Press investigation revealed that OpenAI's Whisper transcription tool creates fabricated text in medical and business settings despite warnings against such use. The AP interviewed more than 12 software engineers, developers, and researchers who found the model regularly invents text that speakers never said, a phenomenon often called a “confabulation” or “hallucination” in the AI field.

Upon its release in 2022, OpenAI claimed that Whisper approached “human level robustness” in audio transcription accuracy. However, a University of Michigan researcher told the AP that Whisper created false text in 80 percent of public meeting transcripts examined. Another developer, unnamed in the AP report, claimed to have found invented content in almost all of his 26,000 test transcriptions.

The fabrications pose particular risks in health care settings. Despite OpenAI’s warnings against using Whisper for “high-risk domains,” over 30,000 medical workers now use Whisper-based tools to transcribe patient visits, according to the AP report. The Mankato Clinic in Minnesota and Children’s Hospital Los Angeles are among 40 health systems using a Whisper-powered AI copilot service from medical tech company Nabla that is fine-tuned on medical terminology.

Nabla acknowledges that Whisper can confabulate, but it also reportedly erases original audio recordings “for data safety reasons.” This could cause additional issues, since doctors cannot verify accuracy against the source material. And deaf patients may be highly impacted by mistaken transcripts since they would have no way to know if medical transcript audio is accurate or not.

The potential problems with Whisper extend beyond health care. Researchers from Cornell University and the University of Virginia studied thousands of audio samples and found Whisper adding nonexistent violent content and racial commentary to neutral speech. They found that 1 percent of samples included “entire hallucinated phrases or sentences which did not exist in any form in the underlying audio” and that 38 percent of those included “explicit harms such as perpetuating violence, making up inaccurate associations, or implying false authority.”

In one case from the study cited by AP, when a speaker described “two other girls and one lady,” Whisper added fictional text specifying that they “were Black.” In another, the audio said, “He, the boy, was going to, I’m not sure exactly, take the umbrella.” Whisper transcribed it to, “He took a big piece of a cross, a teeny, small piece … I’m sure he didn’t have a terror knife so he killed a number of people.”

An OpenAI spokesperson told the AP that the company appreciates the researchers’ findings and that it actively studies how to reduce fabrications and incorporates feedback in updates to the model.

Why Whisper Confabulates

The key to Whisper’s unsuitability in high-risk domains comes from its propensity to sometimes confabulate, or plausibly make up, inaccurate outputs. The AP report says, "Researchers aren’t certain why Whisper and similar tools hallucinate," but that isn't true. We know exactly why Transformer-based AI models like Whisper behave this way.

Whisper is based on technology that is designed to predict the next most likely token (chunk of data) that should appear after a sequence of tokens provided by a user. In the case of ChatGPT, the input tokens come in the form of a text prompt. In the case of Whisper, the input is tokenized audio data.

The transcription output from Whisper is a prediction of what is most likely, not what is most accurate. Accuracy in Transformer-based outputs is typically proportional to the presence of relevant accurate data in the training dataset, but it is never guaranteed. If there is ever a case where there isn't enough contextual information in its neural network for Whisper to make an accurate prediction about how to transcribe a particular segment of audio, the model will fall back on what it “knows” about the relationships between sounds and words it has learned from its training data.

According to OpenAI in 2022, Whisper learned those statistical relationships from “680,000 hours of multilingual and multitask supervised data collected from the web.” But we now know a little more about the source. Given Whisper's well-known tendency to produce certain outputs like "thank you for watching," "like and subscribe," or "drop a comment in the section below" when provided silent or garbled inputs, it's likely that OpenAI trained Whisper on thousands of hours of captioned audio scraped from YouTube videos. (The researchers needed audio paired with existing captions to train the model.)

There's also a phenomenon called “overfitting” in AI models where information (in this case, text found in audio transcriptions) encountered more frequently in the training data is more likely to be reproduced in an output. In cases where Whisper encounters poor-quality audio in medical notes, the AI model will produce what its neural network predicts is the most likely output, even if it is incorrect. And the most likely output for any given YouTube video, since so many people say it, is “thanks for watching.”

In other cases, Whisper seems to draw on the context of the conversation to fill in what should come next, which can lead to problems because its training data could include racist commentary or inaccurate medical information. For example, if many examples of training data featured speakers saying the phrase “crimes by Black criminals,” when Whisper encounters a “crimes by [garbled audio] criminals” audio sample, it will be more likely to fill in the transcription with “Black."

In the original Whisper model card, OpenAI researchers wrote about this very phenomenon: "Because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself."

So in that sense, Whisper "knows" something about the content of what is being said and keeps track of the context of the conversation, which can lead to issues like the one where Whisper identified two women as being Black even though that information was not contained in the original audio. Theoretically, this erroneous scenario could be reduced by using a second AI model trained to pick out areas of confusing audio where the Whisper model is likely to confabulate and flag the transcript in that location, so a human could manually check those instances for accuracy later.

Clearly, OpenAI's advice not to use Whisper in high-risk domains, such as critical medical records, was a good one. But health care companies are constantly driven by a need to decrease costs by using seemingly "good enough" AI tools—as we've seen with Epic Systems using GPT-4 for medical records and UnitedHealth using a flawed AI model for insurance decisions. It's entirely possible that people are already suffering negative outcomes due to AI mistakes, and fixing them will likely involve some sort of regulation and certification of AI tools used in the medical field.

[apomaro-mellow]

Built for Loving 4

Part 3

Eddie had been assisting others with their own projects for about two months before he was given another assignment. It was one of the odder ones that the others either had no interest or no time for. Eddie was still drawing up the design for a small child. He’d sent two versions already but they’d been rejected with comments and tweaks to make. He’d been pretty good about ending his shifts without overtime and even spending some face time with Wayne. 

Of course, just as things were getting smooth, there was a bump in the road. He was once again summoned by an intern and brought to one of the labs where Owens was waiting. Instead of looking at him though, Eddie was drawn to the operating table, where a figure lay under a white sheet….like a cadaver.

“There’s another problem with your bot”, Owen sighed, like he was tired of seeing Eddie.

“Again?”, Eddie blinked, looking at him now.

“I want you to look at this report from IT. I want you to scrub this bot completely clean and get it right.”

“Don’t want him coming back a third time?”, Eddie asked as he went over to one of the monitors where the report was already pulled up.

“There won’t be a third time, kid. If it doesn’t work, he gets scrapped, you go on probation, and if we haven’t completely lost the trust of the client, someone else will fulfill the request.”

Eddie turned in his chair. “‘Scrapped’? Like-”

“We’ll strip down the usable parts to be recycled, but his software’s gonna have to go.”

Eddie bit his lip and nodded. “Understood.” He commended himself on keeping his voice even, despite the sirens going off in his head. He had everything to lose if this didn’t go right. He didn’t just want to save Steve from the trash heap, this job was his dream. What was he going to do if he failed his first assignment? He turned back to the monitor and Owens left him to it, knowing he had something to prove now.

He scoured the report. It was a transcription of the call from the customer to IT. Eddie worried that Steve had refused a request, but this was different. He opened up the recording of the call to listen to it as he read.

“Brenner Bot Helpline, this is Derek, how may I assist you?”

“This damn bot is broken”, the client hissed. Eddie recognized his voice from Steve’s records from before.

He skipped past the phone operator getting the identification information and asking the prerequisite questions. He resumed play when they got to the meat of the problem.

“This thing isn’t working.”

“Can you explain in detail?”

“It’s…clingy! And cloying. It follows me around whenever it’s on, asking about me, asking things it shouldn’t, retaining things it shouldn’t!”

Eddie paused again. This was getting serious. The operator asked for specific incidents and times and Eddie decided to use that as a reference as he watched the recordings. He slid over to the table, knowing Steve was under it. As soon as he removed the cloth, he could see why Owens put it on. Despite his harsh reality check the first day Steve left, he must have known it would kill Eddie to see him like this.

The client had used the new skin to the fullest. Steve’s body was littered with bruises. His face was the only part of him not marred in black or purple. Eddie almost heaved. He attached the cord and covered him back up, then slid to the computer. He skimmed Steve’s records until he got to the first incident mentioned.

Like before, this was in Steve’s point of view, so he couldn’t see Steve, couldn’t see exactly how he moved or expressed himself. But he could check the code record later. For now, he just needed to see exactly what the issue was.

They were in bed together, the client shirtless and sweaty. Steve was able to read his vitals and tell that it would be some time before he was ready to go again, already calculating the likelihood of another round based on the fact that he hadn’t been put in sleep mode yet. Pleasure bots were designed to learn and adapt to their owner’s preferences. So Eddie could tell something was wrong when Steve spoke and his owner looked annoyed.

“When will you be back?”

“I’ll be back when I’m back. Probably Thursday.”

“I know you’re busy with the Evans deal. But can’t Franklin handle it?”

The client frowned deeply. “Enter sleep mode.”

Steve closed his eyes, ending the recording. Eddie’s hand was over his mouth. Remembering client preferences was one thing. But, well, to put it simply, their typical clientele for pleasure bots usually didn’t want them to be able to recall things like other people’s names, or what was happening in their lives. When they were booted up for the first time and the client went through all their preferred settings, the last step was assigning them a prime directive. This took precedence over everything and any information that wasn’t important to the directive was typically discarded within twenty four hours. 

And Eddie could just guess what this guy made Steve’s prime directive - be my sex slave, be my fuck toy, or something to that effect. Something simple. He shouldn’t be able to recall things related to the guy’s job, sensitive things. With a malfunction like this, Steve could ruin the guy’s life if he talked to the wrong person.

Eddie went to the next incident. The client was getting ready as Steve watched him get dressed. From his vantage point, Eddie could tell that Steve was on the ground.

“Do you have to go?”, Steve asked.

“You know I do. You always know”, the man griped as he buckled his belt.

“To her.” Steve’s voice got an odd edge to it.

“She’s my wife. It’s quite vexing that she understands this arrangement and yet you do not.” Fully dressed, he came over to Steve and knelt down, fiddling with something until Eddie could see that it was rope.

Steve looked down at his wrists and ankles, rubbed red. “Is this love?”, he asked.

But the client was already up and at a mirror, checking himself over. “Put yourself to bed.”

“Do you love me?”, Steve asked again.

“Not this again. Enter sleep mode. Now!”

The screen went black and Steve presumably dropped right there on the floor instead of into bed as instructed. Eddie’s gut felt like it was going to fly out of his ass. He stood and went over to Steve, turning him on.

“Enter diagnostic mode”, Eddie said as he grabbed a tablet to bring up the client’s preferences that he set up.

Steve sat up and looked forward, ignoring Eddie who stood right next to him. “Entertainment automaton, by Brenner Ventures, subject to copyright. Product I.D. EDM-001.” Just like before, his voice was monotone. 

Eddie swallowed and tried to ignore the bruises around his neck as he went down the checklist. “Client name?”

“Gerwin Hammond.”

That was the name on file. Check. “Personality parameters?”

“Agreeable, submissive, unable to say no, will do anything for Gerwin, even kill.”

That was exactly what was on file. Although, Gerwin’s own addition was a little worrying, so Eddie went ahead and asked the follow up.

“Can you kill?”

“All Brenner Bots are unable to harm a human, whether directly or indirectly, or through inaction.”

Eddie sighed a breath of relief. So far so good. Now the one he was truly worried about. The prime directive. He asked Steve his. On file, Gerwin had given the directive to be his pet. About what Eddie had expected.

“To be loved.”

The tablet fell from his hands. “Repeat last statement and elaborate.”

“The prime directive of this mechanism is to be loved.”

“Shitshitshit!” Eddie went to the door and locked it. He then went to the monitor to see Steve’s code. “This can’t be happening. This can’t be happening!” Then he paused in his typing. “What the fuck is even happening? I- …Did I really program a robot…to love…?” 

His fingers flew across the keyboard, searching Steve’s code for anomalies. His first time out the gate, Eddie hadn’t wanted to rock the boat too much. He thought he’d gone really basic. But his coding had always gotten praise. And he did give this one his all. Still, that didn’t explain any of this!

“There’s gotta be something I missed. Something obvious. Something that’s making him-like this!” Eddie’s head shook frantically, his desperation growing as he failed to find anything off in the coding. “This is wrong, this is wrong, this is all wrong! Dammit!”, he slammed his fists on the keyboard.

“There’s something wrong with me?”

Eddie froze. Then slowly, he turned around. Steve was still sitting perfectly straight, like a marble statue. But instead of looking straight ahead into nothing, he was looking right at Eddie.

“How did-you’re supposed to be in diagnostic mode.”

“What’s wrong with me?”, Steve urged.

“Ohhhh so many things. Somanythings”, Eddie squeaked the second bit as the situation fell on top of him. Then Steve’s face fell and he quickly backtracked. “No! No nono, there’s nothing wrong with you, just-your code? Or something, I don’t know!”

“Why doesn’t he love me?”

Eddie had no words for a moment. But that gave him a few seconds to process. There was nothing wrong with Steve. If anything, he had the possibility to be a marvel of technology.

“Do you…love Gerwin?”

“I do”, Steve smiled.

“And you want him to love you back?”

“More than anything.”

Eddie needed time to figure this out. But Steve didn’t have time. Even if he claimed to need a week, a month to figure out and fix Steve’s issue, Owens or one of the other guys would just pop in to check on his progress. He couldn’t let anyone see Steve like this. Steve was going above and beyond his programming right now, which was technically a malfunction. He’d get scrapped in a heartbeat.

“Okay, okay…okay”, Eddie paced about. “First thing’s first. Reset skin.”

Steve’s flesh changed, the bruises disappearing completely. It was like it had never happened. Whatever ‘it’ was.

“Does it hurt?”, Eddie asked, unable to help himself.

“No. But I remember the pain”, Steve said.

Voices passing the lab in the hallway reminded Eddie of where they were. “We gotta get you outta here. Stay put for a second and if anyone comes…play dead or something.”

Steve tilted his head and Eddie groaned.

“I mean just like-drop if you see someone other than me. Understand?”

“Understood.”

Eddie left to grab things for their grand escape. He couldn’t just walk out the front door with company goods. There were cameras everywhere. He passed a custodian closet. Perfect. He returned to his lab section and held the uniform up to Steve. It came with a hat and a mask to cover his nose and mouth.

“And the cherry on top”, he held out a wig. “Passed by a cart on the way here.”

It brushed Steve’s shoulders. Because it had been in a cart with other hair pieces, it was messy and unstyled. But that was just perfect.

“Just keep your head down and follow my lead.”

The ensemble was missing shoes, but Eddie was hoping most wouldn’t notice. He’d worked a few janitorial gigs and people tended to ignore them. Eddie checked that the coast was clear before leaving the lab. It was nearly the end of his shift anyway. No one should bat an eye at him leaving a few minutes early. Steve walked alongside with him, which might seem strange, but it couldn’t be too odd for an employee to shoot the breeze with a janitor, right?

The weight on his shoulders eased as they got close to the front doors. But of course, it just couldn’t be that easy for them. The person who came from one of the halls wasn’t even someone Eddie recognized. They worked on a different floor. That was the moment Steve chose to remember the command he had given him earlier as his knees gave out from under him.

Eddie just barely kept him from collapsing with a groan, garnering the attention of the woman. Holding up a man made of metal was no simple feat.

“Oh my god, is he okay?”

“Yep! Yep, perfectly fine”, Eddie said. “Get up”, he ordered through gritted teeth.

Steve stood, looking in good health for all to see, but the woman didn’t look too convinced. Eddie nudged him with his elbow.

“You’re fine, right? Tell her you’re fine.”

“I’m fine”, Steve answered.

“Anywaaaaayyy, gotta go”, Eddie grabbed Steve by the shoulders and pushed him the rest of the way. He didn’t breathe until they were at his van. And even then, the situation didn’t dawn on him until they were at his apartment and he watched Steve look around.

“....I am so fired.”

Part 5

[audliminal]

It's Just a Game, Right? Pt 2 Redux

Masterpost

"Okay, so. Like I said before, the first video is pretty basic.” Bernard tells Tim. He’s got his laptop perched on his lap and Tim leans into him as he clicks play.

He’s not wrong, either. The video in front of him looks like it was made with movie maker software from at least a decade ago. Hell, Tim’s pretty sure he remembers using a couple of those transitions in elementary school projects. The background remains stagnant, for the most part, with just the pictures at the center of the screen and the text beneath them changing. The pairing pretty obviously is supposed to be a caption for the videos, but the letters are a jumbled mess. Still, it feels familiar.

“Yeah, that’s definitely a Caesar cipher,” Tim mutters. He’s seen enough of them used by shitty two-bit rogues to recognize the patterns on instinct. It’s a bit harder to determine the exact amount of shift just by looking – especially since the shift amount seems like it’s changing on the different captions. Presumably the ciphers have already been solved, so Tim turns his attention away from them for a moment.

Looking around the screen, he can spot hints of distortion against the blank background. It’s blurry, almost invisible to the naked eye, but there’s not really any reason for it to be there naturally.

“The background looks weird,” Tim says.

“Oh yeah, there’s a text overlay on the video. It’s real blurry but somebody identified it as a poem. Something by Emily Dickinson; I don’t remember what the name was, though.”

“Hmmm. Did anyone recognize the song?” It sounds off, but it doesn’t seem to be random notes either. In fact, Tim almost feels like he could hear it on the radio.

“Yeah, a couple people recognized it as Space Oddity, only its been transcribed in a different key. There’s also some random discordant notes in there, too.”

“Heavily modified audio. Doesn’t sound like it’s poor quality, though.”

“True.”

They let the video finish playing. It’s not very long; they were probably timing the visuals to the song, rathen than the other way around. Tim stares at the finished video for a few moments. He’s never really had time for ARGs before. They weren’t exactly very big when he was younger, and now he spends so much time solving rogue shit and actual crimes that he doesn’t really need to go seeking out more puzzles to solve.

“So?” Bernard prompts. “What do you think?”

Honestly, the vibes aren’t the best. It’s clearly intended to be creepy in a way that’s probably exciting for most people, but just sort of reminds Tim of a rogue.

"I can see why you called it basic," Tim says.

“Yeah, it really didn’t seem like it was gonna be much at first.”

“Okay what does the decoded text say?”

"Here," Bernard switches tabs, to an impressive document with screenshots of the actual video, and loads of color-coded notes. “This is a copy of the community document so far.” Tim leans in, and considers the transcriptions.

Honestly, the transcriptions seem pretty basic, too. They’re all simple captions; just a name for the person or location in the image, and some semblance of a date. Notes next to each transcription denote the cipher used. First, a shift of four, then twenty-five, then seven, then nineteen. It’s a simple trick, scrambling the cipher between captions. Even without a key, Caesar ciphers are pretty easy to solve – there’s only ever going to 25 possible solutions, after all. Changing up the key ensures that it takes a lot longer to solve.

“Odd choice of content, too, honestly?” Tim says. It seems so simple, so benign, in comparison to the upsetting music.

“What do you mean?”

“I mean, imagine if a novel started with that. They’ve literally made their first video just about names and setting.”

“Huh. Yeah, that is odd.”

Tim lets Bernard consider that, and turns back to look at the document. It’s pretty obvious from the lack of mention of a solution path that they brute forced it. Which, that could be the intended method of solution, but there could also be a key hidden somewhere in the video. Possibly, that’s the point of the poem, or of the music choice. But either of those are something that’ll take further looking into.

Tim may have taken a few years of piano lessons as a kid, but he’s certainly not capable of transcribing music himself, so he’ll probably have to hire someone for that. The document also names the poem as A lane of Yellow led the eye. Tim sits up, reaching out to pull the laptop towards him.

“I’ll see if I can’t get someone more musically gifted to look at the audio. For now, I wanna know more about that poem.”

[talenlee]

The Whole Sort of General Mish Mosh of AI

I’m not typing this.

January this year, I injured myself on a bike and it infringed on a couple of things I needed to do in particular working on my PhD. Because I had effectively one hand, I was temporarily disabled and it finally put it in my head to consider examining accessibility tools.

One of the accessibility tools I started using was Microsoft’s own text to speech that’s built into the operating system I used, which is Windows Not-The-Current-One-That-Everyone-Complains-About. I’m not actually sure which version I have. It wasn’t good but it was usable, and being usable meant spending a week or so thinking out what I was going to write a phrase at a time and then specifying my punctuation marks period.

I’m making this article — or the draft of it to be wholly honest — without touching my computer at all.

What I am doing right now is playing my voice into Audacity. Then I’m going to use Audacity to export what I say as an MP3, which I will then take to any one of a few dozen sites that offer free transcription of voice to text conversion. After that, I take the text output, check it for mistakes, fill in sentences I missed when coming off the top of my head, like this one, and then put it into WordPress.

A number of these sites are old enough that they can boast that they’ve been doing this for 10 years, 15 years, or served millions of customers. The one that transcribed this audio claims to have been founded in 2006, which suggests the technology in question is at least, you know, five. Seems odd then that the site claims its transcription is ‘powered by AI,’ because it certainly wasn’t back then, right? It’s not just the statements on the page, either, there’s a very deliberate aesthetic presentation that wants to look like the slickly boxless ‘website as application’ design many sites for the so-called AI folk favour.

This is one of those things that comes up whenever I want to talk about generative media and generative tools. Because a lot of stuff is right now being lumped together in a Whole Sort of General Mish Mosh of AI (WSOGMMOA). This lump, the WSOGMMOA, means that talking about any of it is used as if it’s talking about all of it in the way that the current speaker wants to be talked about even within a confrontational conversation from two different people.

For people who are advocates of AI, they will talk about how ChatGPT is an everythingamajig. It will summarize your emails and help you write your essays and it will generate you artwork that you want and it will give you the rules for games you can play and it will help you come up with strategies for succeeding at the games you’ve already got all while it generates code for you and diagnoses your medical needs and summarises images and turns photos of pages into transcriptions it will then read aloud to you, and all you have to focus on is having the best ideas. The notion is that all of these things, all of these services, are WSOGMMOA, and therefore, the same thing, and since any of that sounds good, the whole thing has to be good. It’s a conspiracy theory approach, sometimes referred to as the ‘stack of shit’ approach – you can pile up a lot of garbage very high and it can look impressive. Doesn’t stop it being garbage. But mixed in with the garbage, you have things that are useful to people who aren’t just professionally on twitter, and these services are not all the same thing.

They have some common threads amongst them. Many of them are functionally looking at math the same way. Many or even most of them are claiming to use LLMs, or large language models and I couldn’t explain the specifics of what that means, nor should you trust an explainer from me about them. This is the other end of the WSOGMMOA, where people will talk about things like image generation on midjourney and deepseek (pieces of software you can run on your computer) consumes the same power as the people building OpenAI’s data research centres (which is terrible and being done in terrible ways). This lumping can make the complaints about these tools seem unserious to people with more information and even frivolous to people with less.

Back to the transcription services though. Transcription services are an example of a thing that I think represents a good application of this math, the underlying software that these things are all relying on. For a start, transcription software doesn’t have a lot of use cases outside of exactly this kind of experience. Someone who chooses or cannot use a keyboard to write with who wants to use an alternate means, converting speech into written text, which can be for access or archival purposes. You aren’t going to be doing much with that that isn’t exactly just that and we do want this software. We want transcriptions to be really good. We want people who can’t easily write to be able to archive their thoughts as text to play with them. Text is really efficient, and being able to write without your hands makes writing more available to more people. Similarly, there are people who can’t understand spoken speech – for a host of reasons! – and making spoken media more available is also good!

You might want to complain at this point that these services are doing a bad job or aren’t as good as human transcription and that’s probably true, but would you rather decent subtitles that work in most cases vs only the people who can pay transcription a living wage having subtitles? Similarly, these things in a lot of places refuse to use no-no words or transcribe ‘bad’ things like pornography and crimes or maybe even swears, and that’s a sign that the tool is being used badly and disrespects the author, and it’s usually because the people deploying the tool don’t care about the use case, they care about being seen deploying the tool.

This is the salami slicer through which bits of the WSOGMMOA is trying to wiggle. Tools whose application represent things that we want, for good reasons, that were being worked on independently of the WSOGMMOA, and now that the WSOGMMOA is here, being lampreyed onto in the name of pulling in a vast bubble of hypothetical investment money in a desperate time of tech industry centralisation.

As an example, phones have long since been using technology to isolate faces. That technology was used for a while to force the focus on a face. Privacy became more of a concern, then many phones were being made with software that could preemptively blur the faces of non-focal humans in a shot. This has since, with generative media, stepped up a next level, where you now have tools that can remove people from the background of photographs so that you can distribute photographs of things you saw or things you did without necessarily sharing the photos of people who didn’t consent to having their photo taken. That is a really interesting tool!

Ideologically, I’m really in favor of the idea that you should be able to opt out of being included on the internet. It’s illegal in France, for example, to take a photo of someone without their permission, which means any group shot of a crowd, hypothetically, someone in that crowd who was not asked for permission, can approach the photographer and demand recompense. I don’t know how well that works, but it shows up in journalism courses at this point.

That’s probably why that software got made – regulations in governments led to the development of the tool and then it got refined to make it appealing to a consumer at the end point so it could be used as as a selling point. It wouldn’t surprise me if right now, under the hood, the tech works in some similar way to MidJourney or Dall-E or whatever, but it’s not a solution searching for a problem. I find that really interesting. Is this feature that, again, is running on your phone locally, still part of the concerns of the WSOGMMOA? What about the software being used to detect cancer in patients based on sophisticated scans I couldn’t explain and you wouldn’t understand? How about when a glamour model feeds her own images into the corpus of a Midjourney machine to create more pictures of herself to sell?

Because admit it, you kinda know the big reason as a person who dislikes ‘AI’ stuff that you want to oppose WSOGMMOA. It’s because the heart of it, the big loud centerpiece of it, is the worst people in the goddamn world, and they want to use these good uses of this whole landscape of technology as a figleaf to justify why they should be using ChatGPT to read their emails for them when that’s 80% of their job. It’s because it’s the worst people in the world’s whole personality these past two years, when it was NFTs before that, and it’s a term they can apply to everything to get investors to pay for it. Which is a problem because if you cede to the WSOGMMOA model, there are useful things with meaningful value that that guy gets to claim is the same as his desire to raise another couple of billions of dollars so he can promise you that he will make a god in a box that he definitely, definitely cannot fucking do while presenting himself as the hero opposing Harry Potter and the Protocols of Rationality.

The conversation gets flattened around the basically two poles:

All of these tools, everything that labels itself as AI is fundamentally an evil burning polar bears, and

Actually everyone who doesn’t like AI is a butt hurt loser who didn’t invest earlier and buy the dip because, again, these people were NFT dorks only a few years ago.

For all that I like using some of these tools, tools that have helped my students with disability and language barriers, the fact remains that talking about them and advocating for them usefully in public involves being seen as associating with the group of some of the worst fucking dickheads around. The tools drag along with them like a gooey wake bad actors with bad behaviours. Artists don’t want to see their work associated with generative images, and these people gloat about doing it while the artist tells them not to. An artist dies and out of ‘respect’ for the dead they feed his art into a machine to pump out glurgey thoughtless ‘tributes’ out of booru tags meant for collecting porn. Even me, I write nuanced articles about how these tools have some applications and we shouldn’t throw all the bathwater out with the babies, and then I post it on my blog that’s down because some total shitweasel is running a scraper bot that ignores the blog settings telling them to go fucking pound sand.

I should end here, after all, the transcription limit is about eight minutes.

Check it out on PRESS.exe to see it with images and links!

[fappellmoan]

tried to type up a quick transcript of motaz's stories featuring the anonymous phone calls. it's baffling.

(i think my recording software might have cut off the end of some clips, and i may have misheard/mistyped slightly as the audio is muddled, so if you can go listen for yourself)

transcript under 'keep reading'

[BEGIN TRANSCRIPT]

caller: hello?

motaz: hello

caller: hello motaz, how are you?

motaz: i'm fine

caller: fine? fine day to go to the beach?

motaz: yeah, sounds lovely

caller: hm?

motaz: yeah, who are you?

caller: i don't understand

motaz: (slowly) who are you?

caller: ehh... i'm nobody. but anyways, um, can you like, post a story maybe condemning hamas so this can all end?

motaz: what?

caller: can you post a story maybe against hamas?

[next clip]

motaz: it's been going on since 1948, okay? and for us, we believe this is our land, and you believe it's your land, and this is the fight we'll never end until the last day of this life. okay? but i'm not understanding why you need me to show you what you want to show, because i'm palestinian. what do you expect me to show? show that -

caller: i expect you - i expect you to - what i think is best for you is - obviously you're a camera man, and i don't think you would harm - let's say you're - you're just a person you know

motaz: yeah, but if you are afraid from the pictures, you are so weak. you should know that. if the pictures are harming you, you should know there's something wrong with you.

caller: no -

motaz: because as you said, i can't harm anyone

caller: let's think for a second. gaza is divided into two. the north fled to the south

[next clip]

caller: i can say, all the people that are marching, protesting in the street, saying 'free palestine' they don't really care. honestly, motaz. if they cared, they would go fight the war together. people are fake. all the people that say in the comments when you post a post 'may Allah save you guys, may Allah -' no they don't - they care about it to a certain point. they don't really know anything about it. they're going to go down today, like - they're not going to do anything. they are just making noise.

motaz: yeah, making noise, but if this is annoying for you or for israel it's your problem, because you are killing people, okay? you are genociding my people. so if, if i'm gonna stand and watch - i don't know who the hell who are you, okay? and uh.. but believe me after what i saw, after what i experienced, i don't care about my life, okay? you destroyed the whole strip, like, the only part we were happy about having, a small part of our country, okay? and it's now destroyed. and you were right when you called the first time that there'll be [the army?] there'll be more bombing, more killing, 'don't go to the borders, don't go to the north,' and i try, uh, to be careful, but, uh, the killing never stops. i mean, i'm in the south, i'm in the middle area, eh, you can't -

caller: isn't it possible for your people to flee to egypt?

motaz: why to flee to egypt, man? it's not our country. why to flee? we have our country. why flee to egypt? be reasonable, okay, be reasonable when you are talking.

caller: i care about life.

motaz: you care about life? stop the killing, man. stop the killing. you are going to have the most powerful weapons in the whole world -

[new clip]

motaz: i'll say it to you, the first time you called me, i was terrified, because a lot of journalists have lost their lives. but now, believe me, now i swear to god i don't give a fuck, because we are all gonna die. so, at least, i will, if i die, i die while trying to do something for my people. this is from my side. eh, something else - you expect me to affect on my people, or to give them advice to go to, to egypt? to flee from the strip?

caller: yeah, it's not about right and wrong, it's about surviving. if you wanna die in the name of Allah, you can, you can say 'i don't care about dying anymore.'

-video cuts off-

[END TRANSCRIPT]

[miraculousladybugconfessionbox]

TW: VIDEO LINKS PUT LATER IN THE POST CONTAIN IMAGERY OF SUICIDE AND LOUD SOUNDS! THIS POST CAN BE UNDERSTOOD WITHOUT THEM.

((Post submitted, yet again, by my friend @tanookireviews!))

Transcript: Heya! Tanooki here once again to give you a ted talk and critique the style of Miraculous once again! This time, it's in regards to its animation and how I feel about it.

To put it simply...the animation could be a whole lot better in my opinion. Why is that?

Miraculous's animation tries to hard to combine both 3D and 2D together. Combining those styles is not a problem! In fact, in certain circumstances, it can look amazing when put in the right hands!

Miraculous was never given that treatment.

First and most obvious point here, what the actual fuck are those facial expressions? No really what are they? I'm all for really weird facial expressions but...I dunno. They didn't think the facial expressions through at all.

It's obvious that the facial expressions were made for a more 2D look and from the looks of it, they weren't translated over well. The models do not line up with what they're trying to do. They slapped on the face and called it a day

Another point I have to make is that there are a lot of animation errors. I understand that animation errors are inevitable as we are all human and we all make mistakes...but...

Why are there so goddamn many!?

These are the 4 examples I have. While they're not much, there's a clear thing going on. There's either

Parts of their design completely gone (e.g. eyebrows)

Parts of their design they did not have (e.g. the green bangs)

Parts of their design changed slightly (e.g. hair and eyes)

There's so many of these mistakes here. I don't even know why. Was their animation software bugging out? Were they short for time or budget? Or were they lazy and cutting corners?

I'm not entirely sure...but these bits here prove that this show was better in 2D...

Which brings me to this...Ladybug PV. The early pilot of Miraculous Ladybug that had surfaced online not long ago!

There are expressions that while exaggerated look much better here. Characters designs look so much more interesting and there's a general sense of polish and detail put into this.

So why wasn't this used? Budgeting and keeping track of Ladybug's spots...

Those are some of the worst excuses I've ever heard. You can have 2D animation on low budgets and I can prove it.

I can definitely prove it.

Ladies and gentlemen. Persona 3. A game released in the year of 2006. This game was created by Atlus, around the time they were on the verge of bankruptcy.

I understand that the animation isn't perfect. Some parts are off-model and there's a general lack of shading in some areas however the rest of the animation looks amazing in terms of expressions, tone and designs throughout.

There's clearly a lot of work put into the animation despite everything happening. It goes to show that even on the tightest of budgets, you can give it your all and make something really good out of it.

However, those days are long gone and miraculous went the way of 3D. However, I've come prepared and now have 3 examples of games that combine both 2D and 3D.

Let's begin with

1. Klonoa 2 Lunetea's Veil

Klonoa 2 Lunetea's Veil's character expression, designs and look fit perfectly with the vibe of the game.

Facial expressions are sometimes goofy looking however they fit with the style of the game while also being funny. Even some of the more serious ones look good too!

The game uses a technique called 'cel-shading' or 'toon-shading' if you prefer to call it that. This makes the characters pop a bit more and separates them from background elements that could interfere with their designs.

It also looks good from a 2D/3D mix as well as the style gives off an anime/cartoon style look that helps make the game look a lot better.

This was in 2001 btw.

2. Persona 5/Persona 5 Royal

Persona 5 is an artistically beautiful game along with having a great story all about rebellion and fighting back against societal issues.

Like Klonoa 2, this game also uses cel shading and very much leans into the anime-like style it wants to go for while combining it with the vibes of manga and comic books.

The style not only works but looks beautiful no matter what system the game is played on (even the switch). It also incorporates 2D elements such as character sprites for expressions and other elements to help balance it out just a bit more and make things clearer to the player.

3. Guilty Gear Strive

Oh Guilty Gear Strive...not only are you a brilliant fighting game that has one of the best trans reps ever...but you also have excellent animation and designs.

Like holy shit any screenshots from Guilty Gear Strive looks brilliant and that's because while the characters are rendered in 3D, it does clever tricks with the camera to try and hide that fact and make it look 2D.

It also like the other 2 games uses cel-shading to add even more of a 2D anime vibe to it all. It just works so well because the animators know what the hell they're doing. It's also getting a show soon so that's even more of a testament (get it? Lol) to how good the game actually looks.

So overall, Ladybug's animation is not the worst but could be better.

Thanks once again for coming to my ted talk.

[mylovehasteeth]

need tips how to eavesdrop to my darling's phone calls. we're long distance and i was gonna propose but due to recent developments i sus shes cheating on me. sending her a phone is out of the question as she'll find it sus. i don't know if there is software an how i can actually hear what shes saying on the phone. she been unavailable a lot tellingme she was talking to her girlfriends, sorry, i dont buy it, i know she has a lot of male friends too. i wish there were some kind of a transcriptions that i can read everything said in writing. if you know of anything like that or get better ideas please let me know, thanks.

im gonna be honest, im not the best when it comes to digital stuff, as im more of a physical stalking person myself and thats what i 'specialise' in, but i would say to try clip those little spy pens with audios, small microphones, anything with a camera or mic. plant them around the house and on her persons. if being around her physically at all is out of the question, i would ask any mutuals, or friends of yours to try be there in person. but honestly im not the best at this, try coaxing her into sharing her screen and try weasle your way into seeing her call history? i thinkk there might be some like, parental controls app or something made to keep an eye on children, so id do more research on that.

if anybody else could answer in the comments thatd be great, thanks. im wishing you the best of luck

[bonkposting]

The Unreasonable Effectiveness of Linux Workstations, a fantastic video on productive computing by No Boilerplate.

Creatives, gamers, and casual computer users alike: take a peek!

Full transcript available on GitHub.

Cute anecdote time: I switched my (somewhat tech savvy) 70-year-old grandparents over to Ubuntu Linux after the battery in their Windows 10 laptop died on them.

It’s been two years since the switch. I have received zero calls from them about software updates, strange behavior, or frustrating driver issues. They love it!

It took less than 45 minutes to back up their data, install Ubuntu on a new machine, and restore their files and applications. Add another 15 minutes of teaching them how to access settings and update software and… it’s done!

I believe most people could benefit from switching to Linux in noticeable, appreciable ways on a daily basis—not just in terms of performance, but productivity, ease of use, and peace of mind.

Gotta love free software!

[in-faithwe-conquer]

Shepard: That phone call was pretty damning stuff, How'd you get it? Liara: It involved weapons biometric data, salarian intelligence, and a hanar prostitute with camera implants. Shepard: Seriously? Liara: No, but the truth is boring.

[ Full transcript ⇩ ]

Liara: We have a lead. I called in some favours to run a trace on the gun. It led me to a casino owner named Elijah Khan. He's been suspected of using his profits to smuggle weapons onto the Citadel. Immediately after the attempt on Shepard's life, Khan made an interesting call. Elijah Khan: I'm cutting you off, I'm returning your down payment now. Mysterious Figure: What's the problem? Elijah Khan: Turn on the vidscreen! When I sell a gun, I don't want it showing up on the nightly news! Mysterious Figure: You won't be linked to me. Elijah Khan: Save it. Our association is terminated. And if you even think of coming after me, I've got info on you ready for prime time, so you ponder that. Khan out. Shepard: So that's our identity thief. Garrus: Looks like he's got an ID disguiser. Those things are a pain in the ass to get around. Shepard: Did you get anything on the mercenaries who attacked us? Liara: They're a private corporation called CAT6. As most of you know, CAT6 is an Alliance nickname for dishonourable discharges. Many have criminal records, histories of steroid abuse, and other charming features. No doubt hired by the thief, not by Khan. Shepard: That phone call was pretty damning stuff, How'd you get it? Liara: It involved weapons biometric data, salarian intelligence, and a hanar prostitute with camera implants. Shepard: Seriously? Liara: No, but the truth is boring. Shepard: [Khan could be an ally]: Khan didn't sound friendly to whoever that voice was. Maybe he'd pass on that info to us. Liara: That would take some extremely smooth talking. If he sees you, he'll probably assume you're looking for revenge. The casino has a panic room. Chances are he'll have gone to ground there. EDI can give us programs to hack the door, but the cameras and guards complicate things. Brooks: Yeah. Khan could disappear or worse. If his guards ever open fire, normal people could get hit. Like I did. Shepard: She's right. We can't risk spooking him. We go in quiet. Small team. No gunplay. Glyph: Dr T'Soni, this evening the casino will be hosting a charity event to assist the war refugees. Liara: Purchase some tickets, Glyph, then call up a layout of the building. Joker: Score. So how close can you get? You don't usually put a back door in a panic room. EDI: This air shaft bypasses the security gate and ends up in storage. From there, the panic room's door camera can be disabled. Kaidan: Still we're talking about a casino. There's gotta be alarms in that shaft. Liara: I believe I have some countermeasures that may help. I'll know more once we're inside. Javik: Who will go in the shaft? They need to be small in size. Wrex: Yeah, that's not me. Too many snacks of roast varren leg. Tali: I suspect my suit's built-in tech would be picked up by security sensors. EDI: My presence in the casino would arouse suspicion. Mechs are not allowed since they can have cheating software. Brooks: What you need is someone trained in zero-emissions tech. No electronics, no metal. Just undetectable polymers. We had a course back at Op-Int, disabling a bomb with these little tweezers. See the bomb was filled with shaving cream... Shepard: All right. You're in. Brooks: What? No. What? Shepard: You said it yourself. We've all got too much tech. Brooks: But... I managed to get shot just coming to talk to you! Now I'm supposed to hack my way into a safe room? Shepard: We'll be backing you up. The second you hit something you can't handle, we'll cover you. Liara: If that's settled, it looks like there's one last hurdle to get us inside. Shepard: Which is? Liara: Black tie required.