#ai hype

[mostlysignssomeportents]

The real reason the studios are excited about AI is the same as every stock analyst and CEO who’s considering buying an AI enterprise license: they want to fire workers and reallocate their salaries to their shareholders

The studios fought like hell for the right to fire their writers and replace them with chatbots, but that doesn’t mean that the chatbots could do the writers’ jobs.

Think of the bosses who fired their human switchboard operators and replaced them with automated systems that didn’t solve callers’ problems, but rather, merely satisficed them: rather than satisfying callers, they merely suffice.

Studio bosses didn’t think that AI scriptwriters would produce the next Citizen Kane. Instead, they were betting that once an AI could produce a screenplay that wasn’t completely unwatchable, the financial markets would put pressure on every studio to switch to a slurry of satisficing crap, and that we, the obedient “consumers,” would shrug and accept it.

Despite their mustache-twirling and patrician chiding, the real reason the studios are excited about AI is the same as every stock analyst and CEO who’s considering buying an AI enterprise license: they want to fire workers and reallocate their salaries to their shareholders.

-How the Writers Guild sunk AI's ship: No one's gonna buy enterprise AI licenses if they can't fire their workers

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

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

[aiweirdness]

Remember seeing something about GPT-4 doing well on standardized tests? It turns out it may have memorized the answers.

other examples that look like memorization

[charliejaneanders]

Resumes with names distinct to Black Americans were the least likely to be ranked as the TOP CANDIDATE for a financial analyst role, compared to resumes with names associated with other races and ethnicities.

OPENAI’S GPT IS A RECRUITER’S DREAM TOOL. TESTS SHOW THERE’S RACIAL BIAS

[from-mars-to-venus]

Here's an Anti-AI-Hype Reading List!

The AI Hype Cycle Is Distracting Companies

AI hype is 'driven by marketing' right now, says former FDIC Chief Innovation Officer

ChatGPT, AI Technology Is Silicon Valley's Last Hope to Avoid Collapse

You Can’t Trust the AI Hype

Hiltzik: AI hype is as old as AI itself - Los Angeles Times

Don't believe the hype: why ChatGPT is not the “holy grail” of AI research | Salon.com

On NYT Magazine on AI: Resist the Urge to be Impressed | by Emily M. Bender | Medium

AI chatbots are having their "tulip mania" moment | Salon.com

For anyone feeling confused and overwhelmed by the most grating hype-cycle this side of NFTs! Share it with your friends, your mom and pop, your dog! Throw it at every redditor you can! Voila!

[alanshemper]

[tangerinebonfire]

There's a Tumblr meme I can't find that discusses the phenomenon where a captive animal like a bird will treat a shoddy fake bird made of felt just like another bird because it can't tell the difference. The meme points out that humans are probably susceptible to this, too, and I think the thing we're now calling AI confirms this.

[ibboard]

"Data analysis and machine learning" versus "AI"

There's a trend to call everything AI these days. And while AI is technically a superset of Machine Learning, it's misleading because people think that there has been a huge step forwards in "intelligent machines".

There hasn't. There have been companies throwing large amounts of text and very complex models to train them to know which words are likely to come next (using huge amounts of electricity in the process, and brushing aside concerns about copyright and ownership). But that's about it.

It's hype. It's bullshit. And it's being over-applied to build excitement and get funding.

So, I'm cross-posting a Mastodon thread that I made:

Today in "this should probably have been labelled data analysis or machine learning": > AI reveals prostate cancer is not just one disease "They identified two distinct cancer groups among these patients using an AI technique called neural networks." - so, they did data analysis and categorisation 🧐 https://www.manchester.ac.uk/discover/news/artificial-intelligence-reveals-prostate-cancer-is-not-just-one-disease/

(toot)

"But it is only now, with advancements in artificial intelligence, that we have been able to show that there are actually two different subtypes at play" NEURAL NETWORKS ARE OLD TECH! It must be over a decade since deep neural networks really took off. This isn't some new "we can do awesome things now that gen AI came along" advancement 🙄

(toot)

Also, do the scientists not get annoyed at losing credit? It's not "AI discovered" any more than it used to be "calculator discovered". The algorithm did the analysis but the human turned data into knowledge to make the discovery. Did they say "scalpel discovers human anatomy"? No, because the scalpel was just a tool that the early surgeon-scientists were using!

(toot)

New rule: until the computer is processing the data by itself, picking the right algorithm and accurately explaining the novel results without human intervention then "AI" didn't discover ANYTHING!

(toot)

It probably shouldn't bother me as much as it does. But I used to work with a fantastic data science team. We're talking "specifically invited by large organisations and government to be a keynote speaker" levels of people. And we did some fun and interesting machine learning work. But we never once pretended that it was anything more than data analysis, number crunching and finding hidden patterns.

So, every time you read "AI does X" then remember that it's almost certainly just "scientists using data analysis did X" or "scientists using machine learning models that have been in use for a decade did X".

[teledyn]

If people start thinking that Alexa, or something like that, deserves any kind of respect, that works to Amazon’s advantage – that’s something that Amazon would try and amplify. Any corporation, they’re going to try and make you think that a product is a person, because you are going to interact with a person in a certain way, and they benefit from that. So, this is a vulnerability in human psychology which corporations are really trying to exploit. Ted Chiang, on AI hype

News Briefs 14-11-2023 - The Daily Grail

[tnlnyc]

Beware the AI Hype

Every day, claims of an “extinction-level threat” by new generative AIs fill the press. Beware the AI Hype as it comes with a carefully crafted sleight of hand. Current State of AI Generative AI has some basic uses around remixing but gets everything else pretty wrong I’m not opposed to generative AI. MidJourney has created the picture at the top of this page. I have edited the content of this…

[ludwinas]

One potential tonic against this fallacy is to follow an Italian MP's suggestion and replace "AI" with "SALAMI" ("Systematic Approaches to Learning Algorithms and Machine Inferences"). It's a lot easier to keep a clear head when someone asks you, "Is this SALAMI intelligent? Can this SALAMI write a novel? Does this SALAMI deserve human rights?"

from Cory Doctorow’s - Pluralistic: The AI hype bubble is the new crypto hype bubble

[midnights-dragon]

don’t let the critics trick y’all on this one bc the marvels is so fucking good I’m so serious. “oh it’s rushed” “oh it’s not serious enough” “oh the villain is cookie cutter” you know what. is it not enough that movies can be fun anymore

[mostlysignssomeportents]

AI ethics vs AI "safety"

They are hemorrhaging a river of cash, but that river’s source is an ocean-sized reservoir of even more cash.

To keep that reservoir full, the AI industry needs to convince fresh rounds of “investors” to give them hundreds of billions of dollars on the promise of a multi-trillion-dollar payoff.

That’s where the “AI Safety” story comes in. You know, the tech bros who run around with flashlights under their chins, intoning “ayyyyyy eyeeeee,” and warning us that their plausible sentence generators are only days away from becoming conscious and converting us all into paperclips.

It’s pure criti-hype: “Our technology is so powerful that it endangers the human race, which is why you should both invest in it and use it to replace all of your workers.”

This form of criticism is entirely distinct from the legitimate realm of “AI ethics,” whose emphasis is on how bad AI is at the things that will supposedly generate those promised trillions. Things like bias, low-quality training data, training data attacks, data ordering attacks, adversarial examples, the endless stream of confident lies, and the high degree of supervision they necessitate.

Add to that the exploitative labor pipeline, the environmental damage, and the public safety risks and a very different critique emerges —one that’s grounded in AI’s shortcomings, not the supposed risks arising from its incredible power.

-How the Writers Guild sunk AI's ship: No one's gonna buy enterprise AI licenses if they can't fire their workers

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

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

[webawee]

this is why you don't fucking trust chatgpt.

[screecherofthenight]

Guys. GUYS. I was thinking about a Wolf 359 social media au and I came to an earth shattering realization. Kepler is a story time youtuber. he’s a fucking STORYTIME YOUTUBER. I’m having a moment.

[from-mars-to-venus]

I think a lot of people overlook the simple fact that AI art's percieved longevity only goes as far as grifters think they can incorporate it into their grift. If you're an actual artist, generative images and text is largely antithetical to the creative process and ergo is not going to be a frequent tool of many artists. If you're an uncreative person, generative tools are going to, at best, be a fun toy to periodically pick up and put down. For grifters, it's an opportunity to try and come up with the most ghoulishly half-assed children's book you can pump out, or something similarly bland of the like. The bottom line is: no lasting value.

Generative models are expensive to build, more expensive still to train and re-train, and they haven't created sustainable added market value for basically anyone working with them right now. OpenAI is nosediving its way to bankruptcy. BING, who is certainly pushing AI art with DALL-E, has seen their market value slightly *shrink* this year. I think the quality decay of Stable Diffusion's output over the last few months is testament to what we can expect for even the best generators in the game right now. Silicon Valley is not primed for creating anything of lasting value for anyone barring, like, the most porn-addicted twenty-three year olds alive. Generative “art” is a hell of a glass canon and it will remain culturally relevant only for as long as our attention spans are fixated on it, which is a crushing Achilles Heel to have right now.

[alanshemper]