#ai software generation

[ice-d-greentails]

How the Devin AI software developer 'might' be a scam

saying that an artificial intelligence is autonimous is one thing, showing that it is indeed the case is another.

the closing line of the video tells viewers to sent in tasks, so Devin can handle those tasks unseen and send you the results.

okay... but if it really is doing all of this in time, why not just live stream the process at work?

and if deep learning is being used, you should have results you can use, that show the steps the ai took at each decision, a la a teacher asking a student to show their working, so that they can see that they didn't use a cheat sheet, or in this case, that a human operator on the other end didn't do the work for them.

they showed us a prerecorded video of the problem being solved, by now we should know that through the wonders of television, time can seemingly skip ahead by editing out the time between.

they give no explanation as to how Devin solves the issues, they only state that ai can do that you know, which holds as much weight as an illusionist saying it's just magic.

until we can see Devin being given a task and solving it in real time, and a results screen to show the path of deep learning it used, Devin is about as believable an ai software as any sci-fi movie robot.

[aiweirdness]

“Slopsquatting” in a nutshell:

1. LLM-generated code tries to run code from online software packages. Which is normal, that’s how you get math packages and stuff but

2. The packages don’t exist. Which would normally cause an error but

3. Nefarious people have made malware under the package names that LLMs make up most often. So

4. Now the LLM code points to malware.

https://www.theregister.com/2025/04/12/ai_code_suggestions_sabotage_supply_chain/

[frameacloud]

Some ways to avoid generative AI in web searches

Our web searches are being filled with more and more AI-generated slop. GenAI creates misinformation that can be difficult to tell apart from the correct answer to whatever question you're looking up. GenAI mixes things up that have similar names. In one example that is especially easy to recognize, genAI has thought that a baseball team and the bird that it was named after must both be the same creature, with a bizarre combination of their eating habits and behaviors. This mistake still happens with specialized topics that could be harder for you to recognize unless if you're already very familiar with the topic in question. GenAI also makes up fake sources and facts out of thin air, and you can't tell until you try to find them somewhere else. The tech is more like advanced predictive text than something capable of research or reasoning, even though it can look enough like it that it can fool you if you don't know how to spot its mistakes.

Not only are the results of genAI unreliable, the source texts and images that they’re based on were used without permission. Sometimes they aren't modified enough to avoid plagiarism, but it's worse than that. Source images have turned out to be private medical photos and intimate personal photos that hackers had stolen and leaked to harass the people in the photos. Another reason why this technology is unethical is that each genAI query has such a high energy cost that it's significantly harmful to the environment, contributing to the climate crisis.

Here are some things that you can do when you use Google, DuckDuckGo, or other conventional general web search engines:

To turn off Google's AI Overview, set "web" as default. Here's how to do that on your devices and web browsers.

Add this string to any web search to only show results from before the genAI fad. Before:2021

Install the web browser extension uBlacklist. You give it a list of web addresses to not show you in your web searches anymore. Other people maintain lists for it that you can subscribe to so that you won’t see certain types of results in your web searches. Follow the instructions in Laylavish’s Huge AI Blocklist to subscribe to that list which will rid your web search results of AI-generated pages or images.

Since genAI slop is getting to be such a big problem on them, use alternatives to conventional general web search engines some of the time:

Use a specialized search engine instead of a general one. For example, if you only wanted to find a particular science article, there are specialized search engines that only look for those. The blog post "Skip Google for Research" has a list of specialized search engines for academics.

GenAI is notoriously bad at math because that’s too far outside the scope it was designed for: advanced predictive text. In any case, genAI is overpowered for math that is easy for computers to do. If you want to use a very advanced calculator, or even ask a math question in natural language, use Wolfram Alpha.

For questions about how to do things, look them up in WikiHow, the Youtube channel Dad, How Do I? or The Ultimate Manuals Library.

You can search within Wikipedia, but unfortunately vandals have been putting machine-generated falsehoods into it, as well as genAI images. To avoid this, use the article history to view versions of the article from before 2022.

Ask yourself if there are some topics that you often do a web search for just because you keep forgetting an answer to something. Start saving those answers in a book or file that you can refer to offline. This is called a commonplace book. For an example of one, Beth and Angel made theirs into a zine, Stuff I Often Google.

How about a completely different sort of web search than present-day Google or DuckDuckGo? Marginalia Search only brings up results that are text-heavy and similar to the web of the 1990s and 2000s. Its software is independent and open-source.

Have you been using a web search engine to take you to the websites that you visit on a regular basis? Switch to saving them in your web browser's bookmarks folder. Your web browser, Firefox, can sync your bookmarks to your other devices. If you often need to refer to a large number of static web pages (ones that don't change what is on them every day), then you can save and organize them in your Zotero, a bibliography management program. You can sort them with folders and tags so you can find them again.

[cyle]

still confused how to make any of these LLMs useful to me.

while my daughter was napping, i downloaded lm studio and got a dozen of the most popular open source LLMs running on my PC, and they work great with very low latency, but i can't come up with anything to do with them but make boring toy scripts to do stupid shit.

as a test, i fed deepseek r1, llama 3.2, and mistral-small a big spreadsheet of data we've been collecting about my newborn daughter (all of this locally, not transmitting anything off my computer, because i don't want anybody with that data except, y'know, doctors) to see how it compared with several real doctors' advice and prognoses. all of the LLMs suggestions were between generically correct and hilariously wrong. alarmingly wrong in some cases, but usually ending with the suggestion to "consult a medical professional" -- yeah, duh. pretty much no better than old school unreliable WebMD.

then i tried doing some prompt engineering to punch up some of my writing, and everything ended up sounding like it was written by an LLM. i don't get why anybody wants this. i can tell that LLM feel, and i think a lot of people can now, given the horrible sales emails i get every day that sound like they were "punched up" by an LLM. it's got a stink to it. maybe we'll all get used to it; i bet most non-tech people have no clue.

i may write a small script to try to tag some of my blogs' posts for me, because i'm really bad at doing so, but i have very little faith in the open source vision LLMs' ability to classify images. it'll probably not work how i hope. that still feels like something you gotta pay for to get good results.

all of this keeps making me think of ffmpeg. a super cool, tiny, useful program that is very extensible and great at performing a certain task: transcoding media. it used to be horribly annoying to transcode media, and then ffmpeg came along and made it all stupidly simple overnight, but nobody noticed. there was no industry bubble around it.

LLMs feel like they're competing for a space that ubiquitous and useful that we'll take for granted today like ffmpeg. they just haven't fully grasped and appreciated that smallness yet. there isn't money to be made here.

[pseudophan]

i know it would only be an unnecessary pain in the ass for him if i did but it does bother me not to respond like THANK YOU SO MUCH to my gp's messages to me through the healthcare online portal thing. he has a tendency to sit and answer messages and prescription refill requests late as hell in the evening because that's when he finally has the time and it feels so wrong to not reply even if it's entirely unnecessary and he neither expects nor wants that

[relaxedstyles]

[spacefruitpress]

For any creators who use Canva but had no idea that they'd been opted in having their work scraped for AI, you can change this in the privacy settings.

Go to Account Details > Settings > Privacy Settings, and choose "off".

This is not switched on across all markets, but it's worth checking for anyone using Canva.

[ottopilot-sfw]

Vibe coding something. This shit is supposed to take my job? All y'all are fucked. We are all fucked.

[lurkiestvoid]

I still think Mass Effect's differentiating approach between "Artificial Intelligence" (true sentient sapient learning growing digital Intelligence) and "Virtual Intelligence" ('fake' intelligence, an Advanced Software, a glorified interactive database interface-slash-personal assistant, limited functionality) would be very useful in current AI discourse because it's correct that we do not have AI. We have VI. The problems come from pretending VI is the same as AI, insisting it's the same as AI, expecting it to perform like AI, and expecting it to understand like AI. but I'm just hollerin at walls

[smotherstories]

What are other people using to create ai images these days? Something is going on with Bing and my standard prompts (which have worked in the past) are not producing any usable images.

[shocotate]

DALL·E 3 image

[cyle]

I know you're on paternity leave so feel free to ignore this if you don't want to think about it, but has there been any progress on open-sourcing Tumblr's front-end? Inquiring minds would like to know

i hadn’t seen any progress on it before i left. there’s a strong willingness to do it, it’s just a big task to get it open-source-able in a sustainable way. a lot of our CI/CD processes rely on stuff that would need to be rebuilt from scratch, i think. totally doable, just not a priority.

but maybe there’s been progress since i left, i dunno! 🤞

[jessciawilson]

Google Veo3 Prompt Builder – Create Hyper-Realistic AI Videos with Ease

Google Veo3 Prompt Builder:

Easily Produce Stunning AI Videos Are you having trouble producing beautiful, polished videos without investing a lot of money in pricey creators or hours of editing? The revolutionary answer you have been looking for is the Google Veo3 Prompt Builder. Regardless of your background—content creator, marketer, educator, or entrepreneur—this ground-breaking tool enables you to create incredibly lifelike videos with ease, even if you have no prior video editing knowledge.

The Problem with Traditional Video Creation Traditionally, producing high-quality videos has required expensive equipment, difficult learning curves, and complicated tools. It is not only time-consuming for busy marketers, solo producers, and small business owners—it is overwhelming.

How the Google Veo3 Prompt Builder Solves This With the Google Veo3 Prompt Builder, making videos is as easy as filling in the blanks. Without knowing any coding or scripts, users can create cinematic, animated, and explainer-style movies with ease with this AI-powered tool, which is completely compatible with Google Veo3.

The Google Veo3 Prompt Builder:

Why Creators Love the Google Veo3 Prompt Builder

✅ Easy-to-Use AI Video Creation Tool This constructor eliminates uncertainty in AI video production. It streamlines the procedure into three steps—fill, paste, and create—and was designed with usability in mind. No technical expertise is needed!

✅ Generate Cinematic Video Prompts Instantly You can select from a variety of pre-made cinematic video prompts, including YouTube videos, instructional films, and marketing movies. Every template is expertly designed to optimize both visual appeal and user engagement.

Create Explainer Videos with Professional Results Do you need to make captivating explanation films for your viewers? Frameworks that assist you are included in the Prompt Builder.

convey concepts in an understandable and eye-catching manner—ideal for SaaS products, course introductions, and more.

✅ Perfect Google Veo3 Tutorial Companion Is Google's video AI new to you? This prompt builder doubles as a covert Google Veo3 instructional. You are never left wondering what to do next because the fill-in-the-blank format guides you through every step of the procedure.

Rebrand, Resell, or Scale with PLR Video Templates This utility comes with a complete PLR license, unlike most others. This implies that there are no limitations on your ability to resell, rebrand, or utilize the content in client projects. For agencies, independent contractors, and educators wishing to expand their content production, it is perfect.

Includes PLR Video Templates for All Niches The Prompt Builder offers editable and infinitely reusable PLR video templates for everything from social media advertisements to product demonstrations. You save time and establish your business as an authority on videos.

Take Control of AI Video Creation Today The Google Veo3 Prompt Builder is a comprehensive shortcut for creating videos, not just a tool. This builder provides all the tools you need to be successful, whether your goal is to educate others how to use Google Veo3, offer customisable templates, or produce explainer videos. Put an end to your frustration and begin producing beautiful videos that inform, entertain, and convert.

[capesandshapes]

Friendly reminder not to partake in nanowrimo via their website or engage with their organization this year.

There's so many alternatives to nanowrimo in so many different fringe groups that are doing their own novel writing challenges in November.

You can also make the decision just to do the challenge and not engage with the organization at all.

But please for the love of God do not partake in nanowrimo and continue to support them now that they have decided to voice their support for generative ai on top of all of their other bullshit as of late.

The truth of the matter is that AI can be an important aid for new writers and old-- dictation and AI proofreading tools can be important assets to so many different people. Provided that you understand the complications and limitations that that assistive ai can have. Not every single dictated word is going to come out perfectly, and there's not a single person in writing today who would tell you that grammarly, prowritingaid, and all the others are any substitute for a real editor. They're only meant to catch the larger errors and make an editor's job easier-- and their datasets have been polluted and become less effective over time.

Assistive AI has its own problems, and can grow worse over time depending on what it is, but it's not what has writers up in arms about nano this year.

Nanowrimo decided to endorse generative ai, which scraps the creative works of thousands of nonconsenting writers and throw it all together in a word soup meant to match up to a paragraph long prompt. They decided to hold on to the 55,000 words goal, knowing that for some people that would represent hundreds of hours of work, and for others it would represent five minutes typing to a chat bot.

There's so much less genuine work and resilience associated with writing a novel via generative ai. Forum posts of people struggling to finish their novels and seeking out support suddenly lose that gleam of human connection and understanding when someone is literally just a post down looking for tech support to help figure out why their automatic writing machine isn't doing the writing thing good.

And all of the ai crap was after it was discovered that a major member in the organization was grooming minors, and exploiting a forum space where young writers were encouraged to seek connection.

Fuck nanowrimo.

Go join one of the many queer, feminist, or poc spaces to write novels this year. There's literally a group for every single niche. Like there's not even just identity related groups, I'm pretty sure there's like a nanowrimo for Greek god retellings this year. Go find one of them and join them instead.

[telekitnetic-art]

cohost has been a strong alternative to get away from ai scraping. they have no ads, no scraping, and a very similar layout to tumblr (with reblogging, tagging, replying, asks, etc).

so far its felt very much like being on tumblr without the toxicity. the biggest loss is that they dont have statistics- but its nice. it makes it feel cozier, and like the people interacting with your content matter more. ive really been enjoying it.

its worth a shot, anyway!

Thanks for the advice, I’ll do some more digging to see if cohost could be an alternative

I’ve seen other people talk about going to Bluesky and tbh while jumping ship feels like a pretty good plan, I’ve been on Tumblr actively for a little over a year now and I don’t think I really have the energy to try and rebuild on a different platform right now. I’ve already gone through a lot of my art and Nightshaded it, so I’ll definitely keep doing that.

Overall, I’m not really sure what my plan is, even after sleeping on it. I’m still really miserable and frustrated after hearing about this bs. What I might do is keep posting art, but heavily nightshade it and maybe glaze it too for good measure. I’ll probably also stop posting WIPs/previews as frequently. Not really sure. Again, all of this is really frustrating and I can’t think of what the future holds for me in terms of sharing my work without feeling miserable.

[dulafer]

AI App for head/face swapping?

Can anyone recommend an App for head and face swapping? Been using FaceApp and Vidnoz with mixed results. Is there one that uses AI?