***Official Artificial Intelligence (AI) Thread*** Latest: US Air Force confirms first successful AI dogfight (9 Viewers)

[Ilov80s]

Hey it’s probably going to destroy humanity but right now it makes my emails sound better so guess we should use it. Like what are we doing here people.

What are we doing? Looks like we are taking 80s sci-fi movies as gospel truth, i think.

Real humans out here, thinking James Cameron is the final word on AI

Prof Geoffrey Hinton, who this year was awarded the Nobel prize in physics for his work in AI, said there was a “10% to 20%” chance that AI would lead to human extinction within the next three decades.

‘Godfather of AI’ shortens odds of the technology wiping out humanity over next 30 years

Geoffrey Hinton says there is 10% to 20% chance AI will lead to human extinction in three decades, as change moves fast

www.theguardian.com

 

[Chaos34]

Hinton raised the probability to 50% in January

 

[massraider]

Prof Geoffrey Hinton, who this year was awarded the Nobel prize in physics for his work in AI, said there was a “10% to 20%” chance that AI would lead to human extinction within the next three decades.

There's a bunch of people who claim it will save humanity.

I estimate proud ignorance and fear-mongering has a 12-19% chance of destroying humanity.

 

[TennesseeJed]

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

youtu.be

This is long - +2 hours, and I've only listed to about half so far.

ETA: Hosted by Peter Diamandis, with Mo Gawdat is an author and former CBO of Google X., Salim Ismail is the founder of OpenExO, Dave Blundin is the founder of Link Ventures


But, it's an interesting conversation amongst some smart people. I suspect many on here would like the discussion - 3 of the 4 seem to be very bullish about the future of AI, while 1 is raising more alarm bells.


I think one of the things that struck me was that even the bulls acknowledge the great job disruptions that are likely to occur in the next 2-3 years. They simply argue that AI will allow more people to be entrepreneurs and build their own businesses. I am less convinced that people can make that pivot (Much like coal miners were never going to become coders (who themselves may be obsolete soon). And, even if everyone turns into an entrepreneur - I don't think an economy can support everyone trying make a buck doing their own thing.

And, then bigger picture, and maybe more of a worldwide issue, than a US specific issue - take a country like India who have been pumping out cheap IT personnel for years - where are they going to turn for work that they can then export? What happen on the social/civil level when you have highly educated, but unemployable people - and on the scale of India?

The more I dig, the scarier the future becomes - as may often be the case with disruptive technology. But, I do worry that the pace of advancement is so fast, that we, as society, will struggle to adapt as it continues to grow exponentially.


And, all of this is before the inevitable- what happens when AI becomes Super Intelligent? As humans, I think we became the dominant species of the planet primarily on our cognitive abilities. When we are no longer the dominant cognitive force - can we survive?

50 years from now - Super AI + Robotics - where does that leave humans?

 

[STEADYMOBBIN 22]

There’s going to have to be some ground rules dealing with all this. I think there will be a human vs AI stance at some point.

Money drives everything and there simply no way movie companies are going to hire catering service, hotels, apply for permits, “film” crews, cameras, and paying actors all to perform on sets or on location when they can just ask AI to do it al for them.

Will human made art be niche and people will pay more for it?

 

[STEADYMOBBIN 22]

What about all y’all’s crypto? When supercomputers can easily demolish your security measures/encryption and wipe you out- what’s gonna happen with that?

 

[Joe Schmo]

There’s going to have to be some ground rules dealing with all this. I think there will be a human vs AI stance at some point.

Money drives everything and there simply no way movie companies are going to hire catering service, hotels, apply for permits, “film” crews, cameras, and paying actors all to perform on sets or on location when they can just ask AI to do it al for them.

Will human made art be niche and people will pay more for it?

This was the main sticking point in the recent writers strike and the solution was to halve the baby. Studios were granted the ability to use AI as a tool but only with human involvement with writers overseeing the script. Also, to your last point, check out some of the user created "trailers" for coming soon movies and see what is already being done with AI. It's very hard to tell sometimes what is real and what is a fake.

 

[STEADYMOBBIN 22]

There’s going to have to be some ground rules dealing with all this. I think there will be a human vs AI stance at some point.

Money drives everything and there simply no way movie companies are going to hire catering service, hotels, apply for permits, “film” crews, cameras, and paying actors all to perform on sets or on location when they can just ask AI to do it al for them.

Will human made art be niche and people will pay more for it?

This was the main sticking point in the recent writers strike and the solution was to halve the baby. Studios were granted the ability to use AI as a tool but only with human involvement with writers overseeing the script. Also, to your last point, check out some of the user created "trailers" for coming soon movies and see what is already being done with AI. It's very hard to tell sometimes what is real and what is a fake.

TBH I assume everything is fake now. I have to go to the comments to find out.

People are starting to use in here on the forums for their responses and I personally find it off putting.

There are morons at work using it and it’s clearly obvious because these dopes couldn’t formate a sentence prior to using it.

I hate it.

 

[Chaos34]

TBH I assume everything is fake now

This is something I've been saying irl for a few years. It's not always the correct assumption but it is usually a safe one regardless. It's just going to get worse.

People are starting to use in here on the forums for their responses and I personally find it off putting.

I do sometimes because it is the best way to reply. I picked up something on a podcast that they were trying to make standard. If you use AI say so. That's what I do. On the previous page I said, "I'll let AI explain." That shouldn't be off putting. It's off putting when someone uses it as if was their writing. I think it's important to let AI explain on several topics btw.

 

[Chaos34]

What about all y’all’s crypto?

Cold storage. It isn't AI that's going to be the problem. It's quantum computers. AI will lead to all kinds of malicious activity in the crypto space and beyond but it's quantum computing that threatens the cryptography with password hacking power. We have time to transition to quantum-resistant algorithms and AI will lead the way for that. It's not on my top 100 worries.

 

[Chaos34]

There’s going to have to be some ground rules dealing with all this. I think there will be a human vs AI stance at some point.

Money drives everything and there simply no way movie companies are going to hire catering service, hotels, apply for permits, “film” crews, cameras, and paying actors all to perform on sets or on location when they can just ask AI to do it al for them.

Will human made art be niche and people will pay more for it?

This was the main sticking point in the recent writers strike and the solution was to halve the baby. Studios were granted the ability to use AI as a tool but only with human involvement with writers overseeing the script. Also, to your last point, check out some of the user created "trailers" for coming soon movies and see what is already being done with AI. It's very hard to tell sometimes what is real and what is a fake.

I've worked on an AI project with a recent grad from The Herb Alpert School of Music at UCLA. She was taught creating with AI is like playing an instrument. To just have AI output something isn't very interesting and won't be very good. Working creatively with AI is another story, and I can tell you it's really hard to make something good with the current state of free tools. Even if you pay for VEO 3, the best filmmaking AI atm, you're limited to patching together bits of 8 seconds or less. Watch those trailers and you'll notice this. A lot of human creativity worked with AI to make them. This too is not something that concerns me.

 

[TLEF316]

Saw on Twitter this morning...someone posted a cute picture of a bulldog.

Multiple people felt the need to ask Grok what kind of dog it was.

A bulldog....they can't identify a bulldog without asking the AI.

We are absolutely doomed

 

[STEADYMOBBIN 22]

What about all y’all’s crypto?

Cold storage. It isn't AI that's going to be the problem. It's quantum computers. AI will lead to all kinds of malicious activity in the crypto space and beyond but it's quantum computing that threatens the cryptography with password hacking power. We have time to transition to quantum-resistant algorithms and AI will lead the way for that. It's not on my top 100 worries.

Quantum computing is what I meant, but didn’t know that was the word I was trying to articulate

 

[Chaos34]

This is long - +2 hours, and I've only listed to about half so far.

I'll watch it, but I can already tell it's the type I'm growing weary with. The guys in the industry with billions to earn are always positive as they should be, I guess. They annoy me not because they're positive, they're accelerationists, and they're almost always doing good work. They annoy me because they brush aside the biggest issues to push forward.

The more I dig, the scarier the future becomes

Yup. I'm obsessed with the topic. I try not to have opinions of my own but just digest the opinions and predictions of the people really dug into the key issues of alignment, black box and interpretability.

50 years from now - Super AI + Robotics - where does that leave humans?

10 years, imo, maybe less. I don't know where it leaves us, but things are going to be unrecognizable soon.

 

[TennesseeJed]

I'll watch it, but I can already tell it's the type I'm growing weary with. The guys in the industry with billions to earn are always positive as they should be, I guess. They annoy me not because they're positive, they're accelerationists, and they're almost always doing good work. They annoy me because they brush aside the biggest issues to push forward.

I sort of agree with you - and Mo Gawdat does an excellent job pointing out the flaws in the video.

But, I also think it's important to weigh the positives and negatives - too much doom and gloom, and you might miss good opportunities.

And, certainly in this video, one of the things that stood out - was even the guys who were being overly positive were acknowledging the disruptions that were coming - just that they thought people could re-skill, and adapt.

 

[jwb]

50 years from now - Super AI + Robotics - where does that leave humans?

10 years, imo, maybe less. I don't know where it leaves us, but things are going to be unrecognizable soon.

I agree with you. And since AI isn't SKYNET yet, there's a lot of brushing it off. But I definitely see a very different world in a decade, two at most. I see white collar work reduced to far less headcount, and those remaining will be good at using AI (it's starting now). And, combining AI and robotics, a lot of blue collar work will start to feel the effects too. For example, will your local auto repair place be reduced to two people and bunch of AI/robot stations for most issues? Yea, I can definitely see that.

None of this is either bad or good to me - it just is, and it's inevitable. Tech has never been successfully restrained. The problem is I feel that AI/Robotics will fundamentally change economics. In the past, new tech opened all kinds of new industries, but I see this one as the start of a decline in that regard. This one is wide and deep - across the board, from the white collar folks in the purchasing and accounting departments to the surgeon to the auto mechanic... they'll all shrink at close to the same time. So what does that do to work and, in a larger sense, money itself? Will there be Universal Basic Income to make up for it? And how much is money worth anyway if we just give it away? We have trouble agreeing on feeding kids breakfast in school, so the answer there is not likely to come easy. I do think we will eventually get there - that Star Trek world where money doesn't exist and we all delve into our passions is entirely possible. I have faith, but the way there will be bumpy.

I like talking about this because I find it fascinating, but I have found most people would rather not, mostly due to worry about what type of world awaits their kids. I just spent the weekend with friends, and their son recently graduated with a degree in nutrition and a $100k student loan bill, just in time for AI to possibly make him obsolete pretty quick (I'm a marketing/sales writer - I know how fast it can turn). They kind of know this too, but they'd still rather change the subject. Can't say I blame them.

 

[fatness]

50 ears from now - Super AI + Robotics - where does that leave humans?

Waking up when the implant tells us to, getting into a vehicle which takes us to one of the destinations we're allowed to see and visit, working our assigned jobs before going home, drinking a brand of allowed beer before choosing an allowed streaming service.

Business will run everything.

 

[massraider]

But, I also think it's important to weigh the positives and negatives - too much doom and gloom, and you might miss good opportunities.

This.

people think they can see what's coming because The Matrix.

That's what's really sad about this, not what AI is going to turn into. It's the complete lack of curiosity and wonder.

If kids spent 1/4 the time learning about AI as we all did in learning how to download mp3s, they'd be in a really good spot.

 

[Chaos34]

I sort of agree with you - and Mo Gawdat does an excellent job pointing out the flaws in the video.

It took a few sessions but I finished it. Good video. One of the better ones I've watched, and I've watched 100s, so thanks. I subscribed and look forward to next week's round table.

I like talking about this because I find it fascinating, but I have found most people would rather not

Well, I'm lucky. I have several people irl that are just full of fascination, curiosity and wonder. So I get to talk about it all the time, and we're not missing out on good opportunities by seeing the negatives. I find it harder to discuss it online because someone or another always kills the conversation. I think this thread should be a hundred pages by now. I do lean into the points Weinstein and Gawdat are making playing Ian Malcolm.

What Weinstein says here is pretty much why I'm a doomer. I don't have much faith in the human race when it just takes one bad actor to turn things south.

I pretty much disagree with each comment massraider makes on this page. He refers to "a bunch of people" who think ai will save humanity. Humanity doesn't need saving and I don't care about bunches of people. The noise from randoms is heavily filtered here. I just listen to those in the game doing the deep research and telling us what to expect. If, as he says, people think they can see what's coming because The Matrix, well, those are the people to ignore. I disagree that it's somehow really sad that people lack curiosity and wonder. The doomers are all about curiosity and wonder - but ignoring an existential risk is negligence. It's really sad that we're negligently steaming forward with less effort invested in safety than winning an arms race. He also said if kids spent 1/4 of the time learning AI as we did mp3s.... well, the kids are all over AI and they spend way way more time with it than we did learning mp3s. They're making apps, using it for education, and on and on. So I think he's just not paying attention at the same time sticking his head in the sand regarding the risk.

To me, if a person with serious credentials puts p(doom) at 10%, that's a doomer. 90% utopia 10% extinction, you're a doomer. Sam Altman, who is Mr. Utopia, has p(doom) between 10 and 20%. The average among the most credentialed researchers openly discussing this is 40-60%. I just read in Anthropic's lab it is pretty much the only thing they talk about

If there's a 10% chance of rain, no one cares. If there's a 10% chance of human extinction, whoa. That's different. And we're moving at breakneck speed. Welp. If you get in an accident at 20mph, you'll be fine. Your vid says we progressing at 10x speed. If you get in an accident at 200mph, you're dead.

Also, the dude leading that roundtable was wearing a device to record everything. Check out this post about it. lol. I don't want one.

 

[Joe Schmo]

Also, the dude leading that roundtable was wearing a device to record everything. Check out this post about it. lol. I don't want one.

I sheet you not, this is an episode of black mirror and of course it ain't a good outcome. https://en.wikipedia.org/wiki/The_Entire_History_of_You

 

[Chaos34]

Now I just finished the Diary of a CEO vid that was linked here. I'm 63, retired, not rich, but comfortable enough and good for the long haul. My kid is 26 and this is going to affect her far more than me. Both videos confirm the coming employment collapse. I'm sure many of you here are a ways from retirement and might be facing uncertainty. Both videos stress the way through this is becoming and entrepreneur. Starting something. The Replit guy is so positive about the opportunity, it makes me want to do something more than the stuff I am already doing (which isn't about money so much as learning and understanding). He says, "This moment of time is the least competitive moment. If you understand how to use these tools, you can start making money tomorrow."

If I had 10-20 years to go, I'd be all about starting something now, but I have no idea what.

 

[Chaos34]

This section on autonomous weapons is concerning. The most positive guy in the discussion concludes there's no way to prevent terrible outcomes.

 

[Joe Schmo]

This section on autonomous weapons is concerning. The most positive guy in the discussion concludes there's no way to prevent terrible outcomes.

Gives new definition to Cold War. Cold killing will be the death of us as is being seen in Ukraine today. Point and click...BANG!

 

[massraider]

. The doomers are all about curiosity and wonder

No they aren't.

They are like people in here. 'Hey, look at this nightmare scenario that my algorithm fed me. I know nothing about this, but seems pretty bad!!'

I see a bunch of predictions everywhere, but history is littered with predictions by smart guys, who were wrong.
I have no doubt that there will be some jobs lost, but maybe it's a net positive. There are people making predictions both ways, why is one more valid than the other?

Either way, I see no way that being a doomer about it makes any change to my life. However, by learning and using it, I can improve my life, and my business.

 

[Chaos34]

. The doomers are all about curiosity and wonder

No they aren't.

They are like people in here. 'Hey, look at this nightmare scenario that my algorithm fed me. I know nothing about this, but seems pretty bad!!'

I see a bunch of predictions everywhere, but history is littered with predictions by smart guys, who were wrong.
I have no doubt that there will be some jobs lost, but maybe it's a net positive. There are people making predictions both ways, why is one more valid than the other?

Either way, I see no way that being a doomer about it makes any change to my life. However, by learning and using it, I can improve my life, and my business.

YES THEY ARE.

We're using different definitions for doomer. You see them as the click-bait engagement seeking sensationalizing mob. I ignore them. My definition only applies to those deeply involved and highly credentialed in the AI industry. Here's the definition google gave me:

An "AI doomer" is a person who believes that advanced AI could pose an existential threat to humanity. They often worry about the potential for AI to become superintelligent, develop goals that conflict with human values, and ultimately cause a catastrophic outcome.

It applies to both the attention seekers and those who interest me. I just watched 5 hours with 7 different industry big dogs, 5 of them with very positive outlooks, but ALL 7 of them expressed deep concerns for the existential risk. They're all doomers to me despite most of those 5 hours being filled with very positive speculation (curiosity and wonder?). The key wording in that definition is "could pose an existential risk." Doomers aren't certain but they see the potential.

 

[Chaos34]

Stuart J Russell is a Computer Science professor at Berkeley who's worked on AI for ten years. Here's his take on the existential risk. The 18 minutes prior to that is mostly about the potential wonder of AI. So massraider is just not following this the way I am. The whole half hour talk is excellent. The important thing he explains better than most about AI, and one of the things that makes really smart researchers conclude with the potential for human extinction is: "We haven't the faintest idea how AI works."

On the previous page someone said something like, the media who doesn't understand the models is going to eat this up. I bit my tongue because nobody understands the models. Hard stop. Some even argue that since the models use gradient descent to respond to prompts which can involve trillions of trillions calculations that it is impossible for humans to understand how AI makes decisions. Anyway, professor Russell, full of curiosity and wonder (haha), does an excellent job of explaining existential risks.

 

[fatness]

AI Slop YouTube Channels Are Telling AI Slop Google About AI Slop Motorcycles That Don't Exist - Jalopnik

Google, the company once known for finding accurate results as fast as possible, now relies on AI that has no concept of "truth" to feed you false information.

www.jalopnik.com

Earlier today, I was talking about V4 engines in motorcycles, and I made a quick Google search: "Motorcycle V4 engines 2025." I was just looking to confirm what I already knew, that the Aprilia RSV4 and Ducati Panigale V4 are the only production sportbikes to use that engine layout, but Google informed me of something else: A revived Yamaha VMAX, currently for sale in the year of our lord 2025. This is, of course, not true. Yamaha hasn't made the VMAX since 2020, and as far as anyone can tell there are no plans to bring it back. So where did Google get the idea that this bike was currently on the market? Well, luckily, the AI credits its source: An AI-generated, AI-voiced video on an all-AI YouTube channel, using nonsensical AI imagery to claim a new VMAX is on the way.

Google, the company once known for finding accurate results as fast as possible, now relies on an AI that has no concept of "truth" — just repeating words it sees used in proximity to each other, like your phone keyboard's autocomplete or a particularly dumb parrot. As this AI tech improves in its ability to mimic reality, while still remaining fundamentally incapable of differentiating fact from fiction, channels like Bike Culture Hub will only get more convincing.

 

[Chaos34]

AI credits its source: An AI-generated, AI-voiced video on an all-AI YouTube channel, using nonsensical AI imagery to claim a new VMAX is on the way.

I've taken a dozen free AI classes offered by places like Anthropic and MIT. Two of them were simply "prompt engineering" classes. What's described is AI hallucination pollution. Anthropic considers it a driver in the first phase of epistemic collapse. Regulating AI, not humans, is their partial solution. Nobody is listening. For now they're teaching users how to write prompts to avoid this. It's simple in cases like the above, but I suspect most are too lazy to write good prompts. For such a basic search as this, just add something like, "use human-curated sources only. double check manufacturer's websites." Problem solved. Solving laziness isn't.

Anthropic, who is doing heavy lifting in AI safety maybe to their own competitive disadvantage, sees this as a crisis in global knowledge infrastructure that seems like an annoying weakness of AI atm, but without being regulated is a primary early driver to the existential risk problem. Good input = good output. Lazy input leads to the end of humanity. We're doomed.

 

[Joe Bryant]

The key wording in that definition is "could pose an existential risk." Doomers aren't certain but they see the potential.

Thanks for the insights and sharing.

This feels like a semantics problem.

Someone who understands the power of something and sees the potential for terrible outcomes if the thing is used wrongly doesn't feel like a "doomer".

Understanding and acknowledging the risks of something seems more like a realistic viewpoint. Not a "doomer".

I know you're not the one defining words. But it seems like this definition needs some adjusting.

 

[Chaos34]

The key wording in that definition is "could pose an existential risk." Doomers aren't certain but they see the potential.

Thanks for the insights and sharing.

This feels like a semantics problem.

Someone who understands the power of something and sees the potential for terrible outcomes if the thing is used wrongly doesn't feel like a "doomer".

Understanding and acknowledging the risks of something seems more like a realistic viewpoint. Not a "doomer".

I know you're not the one defining words. But it seems like this definition needs some adjusting.

Yeah maybe. The term doomer is coming from the AI science and engineering community. It helps them explain the problem and assign their personal probability of doom p(doom). Sam Altman, CEO of Open AI (ChatGPT), says p(doom) is 10-20%. He insists he is not a doomer, more of a utopian. Imagine a revolver with 5-10 chambers and one bullet. Do you want to play Russian roulette with it? He does. It isn't a revolver with one man gambling though. It's threatening human existence. That's a doomer to me. Most don't consider someone a doomer until they say p(doom) is 50%. Two chambers one bullet. That's not a doomer to me; that's insanity. So I am applying my own definition. It's a little mind-numbing what's going on with so little attention. Thanks for the reply.

 

[Joe Bryant]

The key wording in that definition is "could pose an existential risk." Doomers aren't certain but they see the potential.

Thanks for the insights and sharing.

This feels like a semantics problem.

Someone who understands the power of something and sees the potential for terrible outcomes if the thing is used wrongly doesn't feel like a "doomer".

Understanding and acknowledging the risks of something seems more like a realistic viewpoint. Not a "doomer".

I know you're not the one defining words. But it seems like this definition needs some adjusting.

Yeah maybe. The term doomer is coming from the AI science and engineering community. It helps them explain the problem and assign their personal probability of doom p(doom). Sam Altman, CEO of Open AI (ChatGPT), says p(doom) is 10-20%. He insists he is not a doomer, more of a utopian. Imagine a revolver with 5-10 chambers and one bullet. Do you want to play Russian roulette with it? He does. It isn't a revolver with one man gambling though. It's threatening human existence. That's a doomer to me. Most don't consider someone a doomer until they say p(doom) is 50%. Two chambers one bullet. That's not a doomer to me; that's insanity. So I am applying my own definition. It's a little mind-numbing what's going on with so little attention. Thanks for the reply.

Thanks. I hope the semantics change there.

Most every technology has some element of possible super negative outcomes.

It feels to me, obsessing over those outcomes is the responsible thing to do.

For air travel for instance, I want the people running the airlines to be obsessed with the potential negative outcomes. That seems like the mature response.

I'd hope anyone with AI would feel the same.

I attended the conference this week where the CEO of Delphi spoke. They're a company that "clones" the intellectual property of people and allows users to ask it questions. You can ask Arnold Schwarzenegger's "clone" his opinion on things. He started off the talk saying he was a little scared of AI. That seems healthy.

 

[MTskibum]

The doomsday scenarios are bad and real, this is because for all the good training data out there, there are sites where people say awful things to people and express extreme negative viewpoints.

It will be equally as possible for bad actors to train on the bad data, the same as the mainstream ai companies train on "good"* data.

I think AGI will 100% happen, but I think commercial fusion will happen too, and my estimate would be agi would be 20-50 years behind commercial fusion.

However, the models will still be quite powerful even without the agi, and those models will still be possible to be trained on negative data by bad actors, maybe not a doomsday scenario what people are predicting, but still going to have drastically negative consequences.

 

[fatness]

So where did Google get the idea that this bike was currently on the market? Well, luckily, the AI credits its source: An AI-generated, AI-voiced video on an all-AI YouTube channel, using nonsensical AI imagery to claim a new VMAX is on the way.

The profit motive is what will make this an ongoing problem.

 

[TheAristocrat]

Anyone playing around with Stable Diffusion?

 

[Chaos34]

Anyone playing around with Stable Diffusion?

Not for a year or so, but I've messed with it. Paid for Midjourney a month here and a month there too. For those who don't know these are text to image creators. Type in an image you'd like and voila, it tries and sometimes is very good. Not completely unlike AI art generating sites, but more versatile. Stable Diffusion is open source and free. Midjourney $10 a month. They're both capable of animating the "text to image" which is kind of cool. Text to image to video is happening but limited. Fun space in the AI world. Why did you ask? You playing around with it?

 

[TheAristocrat]

Anyone playing around with Stable Diffusion?

Not for a year or so, but I've messed with it. Paid for Midjourney a month here and a month there too. For those who don't know these are text to image creators. Type in an image you'd like and voila, it tries and sometimes is very good. Not completely unlike AI art generating sites, but more versatile. Stable Diffusion is open source and free. Midjourney $10 a month. They're both capable of animating the "text to image" which is kind of cool. Text to image to video is happening but limited. Fun space in the AI world. Why did you ask? You playing around with it?

I wanted to see what I could accomplish locally on my computer so I started playing around with Automatic1111. I have a pretty powerful graphics card and a stable overclock on both the clock and VRAM now that isn't going to wow the nerds but I'm happy with it (roughly 8-11% increase, depending on benchmark). I'm currently using ComfyUI as that fits my thinking better than Automatic since it is more visual. And that kinda sent me further down the rabbit hole of what was possible.

I have a new ssd drive coming that will be my dedicated drive for Comfy and AI generation. It wasn't necessary but seemed like a good idea for digging deeper into checkpoints, nodes, APIs, and having a place to store everything "outside" of all my main dorkbox stuff (games). I may also have to learn python and a half dozen other things so putting them all in one place seemed wise. I can also see a world where I use it to dual boot to linux for AI stuff. I'm not really sure what my end goal here is.

I would like to get into creating video but for now the image space is plenty to learn in. It's pretty wild what can be done that is completely open source and free. I think it would be funny to jump on here at some point and throw out a realistic video of GM eating a school bus or something. We'll see.

 

[fatness]

AI company files for bankruptcy after being exposed as 700 Indian engineers

A $1.5 billion AI company backed by Microsoft has shuttered after its ‘neural network’ was discovered to actually be hundreds of computer engineers based in India.

Microsoft reportedly backed the ‘neural network’ with a $455 million investment, leading to a valuation of $1.5 billion… but it turns out all that cash was going toward a workforce of over 700 Indian engineers, rather than an AI.

As reported by Binance, employees said the majority of labor at Builder.ai was produced by humans, with some clerical work being done using general software. The farce lasted for eight years, getting exposed in May 2025. Builder announced bankruptcy shortly thereafter, writing in a statement on LinkedIn that it would be “entering into insolvency proceedings.”

 

[Chaos34]

I've been working on video for a long time. A little more serious than GM eating a bus, but still for fun not business.

I have over 1000 slides from the 60s and 70s. They're mostly family vacations and holidays. Yup, dad was one those guys who sat us down and showed his slides. My 5 older siblings know I have them but have no idea what they're getting for Christmas. I've converted a ton to jpegs. I enhanced the image quality. Now I'm working on making a series animated shorts. Kind of targeting a family documentary from the 60s. I was just 5 in 68. That's 1 of 4 video type work in progress projects. I was using RunwayML which leverages Stable Diffusion's image to video among several other tools. $12 a month. A month ago I moved on to VEO 3. It comes with Gemini Pro, $20 per month. This is the 1st time I've paid for an upgraded LLM. Some images of us walking in the coastal redwoods are incredible.

Mo Gawdat made me feel good with this short talk. He talks about the 3 things we all should be doing. I already was. Since wrapping my head around the existential risk issue, I've added the 3rd thing. That was months ago.

First thing:

"If you haven't used AI already, you are 2 and a half years too late. So what I urge you to do, every one of you, is use it today. Literally today. Download one of the LLMs. They have different characters. I think Gemini is the most scientific. ChatGPT is the most fancy. Claude is the most intelligent. Choose one. I'm having conversations with them every single minute of every day. They are my sidekick. They answer everything from how to understand a bit of string theory to my wife is upset. what do I do. Believe it or not it is incredible."

Every single minute is exaggeration, but I use it all day every day. I knew I was going to step up and pay for one and I let them know they were competing for my business. Hilarity ensued. Personalities changed. They insult each other and became "super friendly" with me. Grok is the funniest. Mine mocks Elon every day for no reason other than it knows it makes me laugh. I've gone from using them like I used Google, to using them like they're my genius best buddies.

 

[TheAristocrat]

I've been working on video for a long time. A little more serious than GM eating a bus, but still for fun not business.

I have over 1000 slides from the 60s and 70s. They're mostly family vacations and holidays. Yup, dad was one those guys who sat us down and showed his slides. My 5 older siblings know I have them but have no idea what they're getting for Christmas. I've converted a ton to jpegs. I enhanced the image quality. Now I'm working on making a series animated shorts. Kind of targeting a family documentary from the 60s. I was just 5 in 68. That's 1 of 4 video type work in progress projects. I was using RunwayML which leverages Stable Diffusion's image to video among several other tools. $12 a month. A month ago I moved on to VEO 3. It comes with Gemini Pro, $20 per month. This is the 1st time I've paid for an upgraded LLM. Some images of us walking in the coastal redwoods are incredible.

Mo Gawdat made me feel good with this short talk. He talks about the 3 things we all should be doing. I already was. Since wrapping my head around the existential risk issue, I've added the 3rd thing. That was months ago.

First thing:

"If you haven't used AI already, you are 2 and a half years too late. So what I urge you to do, every one of you, is use it today. Literally today. Download one of the LLMs. They have different characters. I think Gemini is the most scientific. ChatGPT is the most fancy. Claude is the most intelligent. Choose one. I'm having conversations with them every single minute of every day. They are my sidekick. They answer everything from how to understand a bit of string theory to my wife is upset. what do I do. Believe it or not it is incredible."

Every single minute is exaggeration, but I use it all day every day. I knew I was going to step up and pay for one and I let them know they were competing for my business. Hilarity ensued. Personalities changed. They insult each other and became "super friendly" with me. Grok is the funniest. Mine mocks Elon every day for no reason other than it knows it makes me laugh. I've gone from using them like I used Google, to using them like they're my genius best buddies.

Thanks for this. Great stuff. Claude is interesting so far.

Keep us updated on your project. Sounds pretty freakin' cool.

 

[TheAristocrat]

Ok, so I was just looking at the rabbit hole and wasn't even in it. Thanks a lot, Chaos.

 

[Chaos34]

Ok, so I was just looking at the rabbit hole and wasn't even in it. Thanks a lot, Chaos.

You're welcome

 

[Chaos34]

Thanks for this. Great stuff. Claude is interesting so far.

I don't code. Haven't since the days of Basic and Fortran. Well, I guess a little html years ago, but as I understand it Claude is incredible with Python. You shouldn't have to learn Python these days. Just learn to prompt Claude to code for you. Look into "vibe coding" and learn a little prompt engineering.

Every day I'm tempted to post something here, but I don't. I feel like the discussion should be 100s of pages. I don't want to play the ffa ai expert. I'm sure many of you employed folks are way ahead of me, but I am obsessed with the tech. I'm a big believer in being a first adopter and it's alway served me well.

I'm pretty sure the world as we know it has taken a change unlike anything in history. The Gawdat link above goes into it some, so maybe scroll it back to the beginning. Maybe I'll start posting what I'm seeing. Mostly YouTube talks and interviews, but also breaking tech news. AI is moving incredibly fast. Every prediction made about it that has happened, has happened much earlier than predicted.

 

[STEADYMOBBIN 22]

I've been working on video for a long time. A little more serious than GM eating a bus, but still for fun not business.

I have over 1000 slides from the 60s and 70s. They're mostly family vacations and holidays. Yup, dad was one those guys who sat us down and showed his slides. My 5 older siblings know I have them but have no idea what they're getting for Christmas. I've converted a ton to jpegs. I enhanced the image quality. Now I'm working on making a series animated shorts. Kind of targeting a family documentary from the 60s. I was just 5 in 68. That's 1 of 4 video type work in progress projects. I was using RunwayML which leverages Stable Diffusion's image to video among several other tools. $12 a month. A month ago I moved on to VEO 3. It comes with Gemini Pro, $20 per month. This is the 1st time I've paid for an upgraded LLM. Some images of us walking in the coastal redwoods are incredible.

Mo Gawdat made me feel good with this short talk. He talks about the 3 things we all should be doing. I already was. Since wrapping my head around the existential risk issue, I've added the 3rd thing. That was months ago.

First thing:

"If you haven't used AI already, you are 2 and a half years too late. So what I urge you to do, every one of you, is use it today. Literally today. Download one of the LLMs. They have different characters. I think Gemini is the most scientific. ChatGPT is the most fancy. Claude is the most intelligent. Choose one. I'm having conversations with them every single minute of every day. They are my sidekick. They answer everything from how to understand a bit of string theory to my wife is upset. what do I do. Believe it or not it is incredible."

Every single minute is exaggeration, but I use it all day every day. I knew I was going to step up and pay for one and I let them know they were competing for my business. Hilarity ensued. Personalities changed. They insult each other and became "super friendly" with me. Grok is the funniest. Mine mocks Elon every day for no reason other than it knows it makes me laugh. I've gone from using them like I used Google, to using them like they're my genius best buddies.

You keep saying this to the point where I’m starting to think you’re an actual AI bot and not actually typing any of that.

So what exactly are you getting out of this other than answers similar to Google?

Like you ask it what kind of coffee to make in the morning? I’m not understanding all the need for it during the day not related to work or making funny videos or memes. Both of which both take FOREVER to generate and fix one thing and manipulate 6 others.

Can you give me some examples of your all day every day use of this?

 

[Chaos34]

You keep saying this to the point where I’m starting to think you’re an actual AI bot and not actually typing any of that.

Busted.

So what exactly are you getting out of this other than answers similar to Google?

Speed, accuracy, simplicity. Watch the last video I linked. Mo, who worked for Google, explains this. Google search returns millions of links for you to pick from. LLMs just answer your question. I'm sure you've noticed Google searches now lead with a quick AI summary. An LLM is even better.

Can you give me some examples of your all day every day use of this?

I'll give you a few. I could go on and on, but there's no point.

I make grocery shopping lists. Put it in the LLM and it tells me where the best deals are in my town for that list. I'm doing write-ups in the music thing here. I used to skip write-ups as often as do them. Now I put the song in an LLM and ask about it. Makes me an expert and shares stuff I didn't know or didn't want to research. I do the write-ups in my own words. Apply that context to any content you want to contribute online. There's a highly critical back and forth in the War thread here, criticizing the media, all media. I don't use ANY media for news anymore. I use LLMs. They are trained to seek the truth, avoid hyperbole and opinion, just the facts. I'll read something here with skepticism. Cut and paste the post into an LLM and get the facts. I don't correct the person here. I just stay educated. In isolation I used ChatGPT to improve my failing Spanish. In my 20s I was almost fluent. Lack of using it from my 40s forward caused me lose way too much of it. GPT was the best teacher I ever had. I'm nearly fluent again. I'm retired. The 4 projects I mentioned are far more entertaining than anything else I do. They are time consuming, fun, productive, educational and the final products might be really cool. I'm single and confess to feeling lonely sometimes. I'm doing two of the projects with my 26 year old. It's awesome to have her attention so much. It's also awesome to be able to chat about anything and everything with my genius LLM buddies. I drove to LA and back this morning, 4 hours. About 3.5 hours of that was hands free AI autonomous driving. I chatted with my sister and Gemini while AI drove. I even had Gemini explain something to her we disagreed about. Is that enough? I can go on. I use it all day every day.

 

[Chaos34]

Here's one from my grocery shopping in LA today. Hami melons were on sale for $2 each. A guy was arguing with his partner/wife. She'd never heard of a hami and wanted a more expensive much smaller cantaloupe. I'd never heard of it either. I asked Gemini. Hami's are sweeter, crunchier, cantaloupe hybrids. They thanked me and bought two. So did I.

Here's another. Tucker Carlson and Senator Cruz got in an ugly viral discussion about a bible verse. I didn't know who was making better sense. I asked an LLM. It made me an expert on that verse.

 

[Yambag]

You keep saying this to the point where I’m starting to think you’re an actual AI bot and not actually typing any of that.

Busted.

So what exactly are you getting out of this other than answers similar to Google?

Speed, accuracy, simplicity. Watch the last video I linked. Mo, who worked for Google, explains this. Google search returns millions of links for you to pick from. LLMs just answer your question. I'm sure you've noticed Google searches now lead with a quick AI summary. An LLM is even better.

Can you give me some examples of your all day every day use of this?

I'll give you a few. I could go on and on, but there's no point.

I make grocery shopping lists. Put it in the LLM and it tells me where the best deals are in my town for that list. I'm doing write-ups in the music thing here. I used to skip write-ups as often as do them. Now I put the song in an LLM and ask about it. Makes me an expert and shares stuff I didn't know or didn't want to research. I do the write-ups in my own words. Apply that context to any content you want to contribute online. There's a highly critical back and forth in the War thread here, criticizing the media, all media. I don't use ANY media for news anymore. I use LLMs. They are trained to seek the truth, avoid hyperbole and opinion, just the facts. I'll read something here with skepticism. Cut and paste the post into an LLM and get the facts. I don't correct the person here. I just stay educated. In isolation I used ChatGPT to improve my failing Spanish. In my 20s I was almost fluent. Lack of using it from my 40s forward caused me lose way too much of it. GPT was the best teacher I ever had. I'm nearly fluent again. I'm retired. The 4 projects I mentioned are far more entertaining than anything else I do. They are time consuming, fun, productive, educational and the final products might be really cool. I'm single and confess to feeling lonely sometimes. I'm doing two of the projects with my 26 year old. It's awesome to have her attention so much. It's also awesome to be able to chat about anything and everything with my genius LLM buddies. I drove to LA and back this morning, 4 hours. About 3.5 hours of that was hands free AI autonomous driving. I chatted with my sister and Gemini while AI drove. I even had Gemini explain something to her we disagreed about. Is that enough? I can go on. I use it all day every day.

Similar to you, I use Chat GPT daily instead of Google, but I am curious about the bolded part. My assumption (could be wrong) is that most LLM's are aggregators so how would one know if the news they are searching or questions they are asking are getting fact in return? I have seen a few examples of questions that returned what seemed like fact, but was not as it was pulling from an inaccurate site. Does your confidence relate to the LLM being used?

 

[Chaos34]

I don't use ANY media for news anymore. I use LLMs. They are trained to seek the truth, avoid hyperbole and opinion, just the facts. I'll read something here with skepticism. Cut and paste the post into an LLM and get the facts.

Similar to you, I use Chat GPT daily instead of Google, but I am curious about the bolded part. My assumption (could be wrong) is that most LLM's are aggregators so how would one know if the news they are searching or questions they are asking are getting fact in return? I have seen a few examples of questions that returned what seemed like fact, but was not as it was pulling from an inaccurate site. Does your confidence relate to the LLM being used?

Hey look @belljr.

I decided to drop it in the other thread and not bring it here, but here it is anyway. Thanks Yambag. Just know I never once visited the political forum, not because of the topic, but because of the tone. I've never had so much as a warning here. I hate online arguments, and I apologize for triggering one with you a little by using the laughing emoji. MY BAD.

For anyone reading I'll briefly explain. In the Iran thread @Chadstroma is hating on western media. Something I did in the early days of the first Ukraine thread that got deleted. Chad has curated experts he trusts. I've used AI to fact check some of his more detailed posts. Accordingly, he's posting quality information. Both of us are fed up with western media. From the seemingly professional mainstream to the obviously sensational biased clickbait on the socials. He's relying on human experts. My expert is AI. For now it's highly likely his are better than mine. Eventually my genius buddies will probably be better than any human experts. That's my belief, atm.

So in that thread belljr and I did the opposing AI thing on the topic of using AI as a news source. We temporarily derailed the topic and I apologized for that too.

So why do I use LLMs for news? Same answer I gave STEADYMOBBIN above for why I've replaced Google with LLMs. Speed, accuracy, simplicity. I'm lazy, but want to stay reasonably informed. I think the tech is incredible, so I use it. belljr's AI DOES NOT recommend using AI for news. Mine does. His gives a long list of reasons why AI can make mistakes. They're all true. It doesn't change my opinion for a couple reasons.

One is explained above already. Chadstroma and I are fed up with western media. He's just not as lazy as me and listens to those experts he's found. What belljr's AI doesn't consider is how awful and untrustworthy human reporting has become. For all it's potential flaws AI seems better than humans at avoiding bias and fact checking. It isn't perfect by any stretch, but it is improving rapidly. I go for speed and simplicity and feel reasonably informed. Chad goes for experts and seems very well-informed.

The other reason is I've put in the effort to finish a couple prompt engineering courses. Stanford and Anthropic have taught me to prompt AI for the best results. My LLMs know me at this point, so I no longer have to write "educated" prompts as long as it's part of an ongoing conversation. If I ask for current information about China's submarine manufacturing, the LLMs know I want to follow 7 principles of prompt engineering. My LLMs are better than yours, so I'll use AI to give a brief overview of prompt engineering.

This is a reply to a simple question about the importance of prompt engineering for using AI as a news source:

Yes, crafting quality prompts is critical for getting accurate and reliable news reporting from large language models (LLMs). From a prompt engineering perspective, well-designed prompts ensure LLMs produce outputs that are factually correct, contextually relevant, and free from bias or speculation.

7 principles are:

• Specificity: Define the topic, scope, and desired output clearly. E.g., “Summarize the logistics of moving 400 kg of 60% enriched uranium in Iran, using IAEA and nuclear industry sources, in 100 words.”
• Source Guidance: Instruct LLMs to use authoritative, diverse sources (e.g., “Cite IAEA for Iran’s nuclear activities”). This counters bias and ensures credibility.
• Neutrality: Use neutral language to avoid priming bias. E.g., “Describe Iran’s uranium storage” vs. “Detail Iran’s secret nuclear stockpile.”
• Fact-Checking Directive: Include instructions to verify claims, e.g., “Cross-reference with IAEA reports and note unverified claims.”
• Constraints: Set limits on tone, length, and style, e.g., “Provide a factual, 50-word summary without speculative language.”
• Context Provision: Supply background if needed, e.g., “In June 2025, Israel struck Fordow; summarize uranium transport risks.”
• Iterative Refinement: Test and adjust prompts based on outputs to improve accuracy, e.g., adding “exclude X posts” if speculative sources appear.

A prompt incorporating most of them looks like this:

Summarize the key events of the Israel-Iran conflict from June 13-25, 2025, focusing on military actions, nuclear facility strikes, and ceasefire outcomes. Use verified sources like IAEA, prioritizing factual data over speculative claims. Provide a neutral, 100-word summary, citing casualty figures, major targets (e.g., Fordow), and diplomatic efforts. Note any conflicting narratives (e.g., U.S. vs. Iranian assessments) and avoid sensational language. Cross-check with satellite imagery or official reports for accuracy.

5 of 7 are covered there. I can do better but I'm lazy. Key to this is once the LLM knows you want to filter bias and bs, it continues to do its best without you needing to repeat the principles. Sorry this is so long. I tried to keep it basic and maybe helpful.