Episode #26 - “Pause AI Or We All Die” Holly Elmore Interview, For Humanity: An AI Safety Podcast

Please Donate Here To Help Promote This Show
www.paypal.com/paypalme/forhu...
FULL INTERVIEW STARTS AT (00:09:55)
In episode #26, host John Sherman and Pause AI US Founder Holly Elmore talk about AI risk. They discuss how AI surprised everyone by advancing so fast, what it’s like for employees at OpenAI working on safety, and why it’s so hard for people to imagine what they can’t imagine.
This podcast is not journalism. But it’s not opinion either. This show simply strings together the existing facts and underscores the unthinkable probable outcome, the end of all life on earth.
For Humanity: An AI Safety Podcast, is the accessible AI Safety Podcast for all humans, no tech background required. Our show focuses solely on the threat of human extinction from AI.
Peabody Award-winning former journalist John Sherman explores the shocking worst-case scenario of artificial intelligence: human extinction. The makers of AI openly admit it their work could kill all humans, in as soon as 2 years. This podcast is solely about the threat of human extinction from AGI. We’ll meet the heroes and villains, explore the issues and ideas, and what you can do to help save humanity.
TIMESTAMPS:
*Progress in Artificial Intelligence (00:00:00)*
Discussion about the rapid progress in AI, its impact on AI safety, and revisiting assumptions.
*Introduction to AI Safety Podcast (00:00:49)*
Introduction to the "For Humanity and AI Safety" podcast, its focus on human extinction threat from AI, and revising AI risk percentages.
*Need for Compute Cap Regulations (00:04:16)*
Discussion about the need for laws to cap compute power used by big AI companies, ethical implications, and the appointment of Paul Christiano to a new AI safety governmental agency.
*Personal Journey into AI Risk Awareness (00:15:26)*
Holly Elmore's personal journey into AI risk awareness, understanding AI risk, humility, and the importance of recognizing unexperienced events' potential impact.
*The Overton Window Shift and Imagination Limitation (00:22:05)*
Discussion on societal reactions to dramatic changes and the challenges of imagining the potential impact of artificial intelligence.
*OpenAI's Approach to AI Safety (00:25:53)*
Discussion on OpenAI's strategy for creating AI, the mindset at OpenAI, and the internal dynamics within the AI safety community.
*The History and Evolution of AI Safety Community (00:41:37)*
Discusses the origins and changes in the AI safety community, engaging the public, and ethical considerations in AI safety decision-making.
*Impact of Technology on Social Change (00:51:47)*
Explores differing perspectives on the role of technology in driving social change, perception of technology, and progress.
*Challenges and Opportunities in AI Adoption (01:02:42)*
Explores the possibility of a third way in AI adoption, the effectiveness of protests, and concerns about AI safety.
Resources:
Azeer Azar+Connor Leahy Podcast
Debating the existential risk of AI, with Connor Leahy
Best Account on Twitter: AI Notkilleveryoneism Memes
/ aisafetymemes
JOIN THE FIGHT, help Pause AI!!!!
Pause AI
Join the Pause AI Weekly Discord Thursdays at 3pm EST
/ discord
/ discord
22 Word Statement from Center for AI Safety
Statement on AI Risk | CAIS
www.safe.ai/work/statement-on...

[@cg000gc]

Americans have the impression that the world stops at their border. Money can move eveywhere and AI can be developed in any country.

[@41-Haiku]

That's why PauseAI specifically pushes for a global treaty. Coordination is difficult, but not impossible! We banned CFCs and blinding laser weapons, and nuclear non-proliferation has worked well enough so far. If consensus about the danger is reached in places of political power around the globe, everyone will save themselves by banning the development of superintelligent AI. I'm not saying that's the most likely outcome, but it is probably the most likely out of all outcomes in which we survive.

[@robxsiq7744]

@@41-Haiku Why would there be a consensus? Japan already said pedal to the metal. many other places aren't as neurotic as the west. The fears are that something bad may happen..thats different than CFCs that you can demonstratably prove something bad always happens when released. This is actually doomsday prophets in tech. Same shit came down about the internet when it was gaining massive momentum around 1998-2000. hackers causing global shutdown of services and even launching nukes.

[@colto2312]

@@41-Haiku then we choose to work, not for profit, but for the sake of work itself. why dig holes in a circle and call it rewarding?

[@PaulAllsopp]

@@41-Haiku While it is a good premise, and many people will adhere to it, do you really think the governments of the world will stop developing AI as a weapon? Those things you mentioned are things developed well after the fact. How long, and what damage did we allow, CFCs to do before banning them? How long have nuclear bombs been around, and why do all the major powers still have them, even though we know they could end humanity? Laser weapons are still being developed and are in use by the US military today.

[@masonlee9109]

@@PaulAllsopp Yes, I think it is possible to get a global treaty, given that AGI looks to be a suicide race. Not easy by any means, but possible. I'd be open to a better idea!

[@goodleshoes]

Let's GO, I'm ready.

[@BrunoPadilhaBlog]

28:48 'They're gonna superly align everything, and it will be fine!' Haha I lolled

[@luciengrondin5802]

As I like to say : we're all running frantically towards a cliff in the hope that we will fly and not fall.

[@ManicMindTrick]

To sharpen that analogy a little, we are running towards the cliff and expect to be able to build an airplane mid air.

[@masonlee9109]

John, Holly, thanks for the conversation; you guys are awesome. Holly, thanks for the work you are doing! I hope to be able to join one of the PauseAI protests soon.

[@BManStan1991]

With how depressed many Americans are nowadays, this may be less of a threat and more of an exciting proposal 😂

[@ForHumanityPodcast]

Lol! Seems to you you would need to be very depressed to want machines to replace humans.

[@41-Haiku]

Jokes aside, I see this a lot in online spaces. I don't think there are a lot of people who want all life to end (though there are some), but a lot of people use big issues like this as a way to express that they are in pain, by using dark humor or being contrarian. As someone who was once in a dark place myself, I genuinely feel for them. Even though in the moment, I often want to say "No, FU, you can't decide for the rest of us that our families don't deserve to live."

[@baraka99]

I'm starting to see the point of accelerationists. Regulatory capture is what a dystopian future is all about, AI in the hands of a few monopolies. Accelerationists want to create enough open source innovation so we have enough strenght and innovation to counter those market forces.

[@GrumpDog]

Exactly.

[@41-Haiku]

Yes, that's a position of many accelerationists, and it is a very serious risk. If I thought we could accelerate and democratize powerful AI systems without extincting humanity, I would be with them! But essentially no one in e/acc is even aware of the technical problem of AI alignment. Creating a powerful AI that does what you want is an engineering problem, and there are dozens of known, concrete, unsolved subproblems. There are clear reasons why it probably goes poorly by default, and it won't go well by magic. The accelerationism dream seems to be that we can wish the future into being, and that bad things won't happen because we presupposed they wouldn't. Regardless, it's not my or PauseAI's position that only a few companies should have this technology. It's that _no one should have it._ Because no one knows what the fuck they are doing.

[@baraka99]

@@41-Haiku Big tech will develop AI either we want or not. Our only change is to level the plainfield.

[@josephboomtv7811]

The more things change the more they stay the same. going to be tons of changes from exponential advances in AI.. lot of those changes will be awesome- and I think the reader will be surprised at what changes do not occur

[@BrianPeiris]

Regarding John's reduced p-doom, and AI plateaus; I'd encourage people to watch Subbarao "Rao" Kambhampati's talks - "Can LLMs Reason & Plan". He's an AI/CS professor at Arizona State University and he makes an excellent case for how we've tricked ourselves into thinking that LLMs are anywhere near AGI. He's subarao2z2 here on youtube. They may well be more breakthroughs in the coming years, but I think it's reasonable to say that getting to true AGI will be a long road.

[@OneRudeBoy]

How do you think AI is a risk rather than just calling AI a risk?

[@41-Haiku]

That's what most machine learning experts say! ("AI Impacts Survey") If you're interested in why they say that, a great starting place is the "AI Safety Info" site.

[@seanmchugh2866]

AI turned me into a newt!

[@rightcheer5096]

I saw most of the Connor Leahy interview and agree with you that it’s borderline confidence-increasing. But it’s just tapping the brakes, and you have to hope the AI industry doesn’t take it as a brake check and cause for running everybody over.

[@uk7769]

technology is not hardware or software or a chip. technology is the application of knowledge.

[@ESTRID_lol]

Hey I was just wondering what the average ai researcher p doom is from? Cheers

[@pratiknaikedu]

Hahaha, oh shit 1:08:00 described me. I will have that UHaul dragging my screens to the woods

[@nunobartolo2908]

The quality data sources have already been used making the models Bigger will just over fit them

[@SimOnTheRoad1]

Ok this might sound wrong because I haven't watched the video yet, but the title clearly says that we should "pause AI or we all die", yet the image is clearly made by AI in order to make people click on it, this is very hypocritical. Nevertheless, the conversation might be interesting and I will listen to it now.

[@uk7769]

all you need to know is what the hyperwealthy class does throughout history. Hint: it is never good for the common folk or the villagers in the way. every human is at risk in this one. we knew this. we ignored the warnings.

[@dogk764]

“OpenAI was founded as a safety org…” good thing irony can’t kill us 😂 (yet, at least)

[@PaulAllsopp]

Funny it's called "Open" AI? It's only been open to Microsoft and nobody else, and we all know Microsoft works hand in hand with humanity...oops, no I mean the military.

[@dogk764]

I think the “open” is more in the context of opening, as in pandora’s box 🤔

[@PaulAllsopp]

@@dogk764 I'm looking forward to testing Grok. While the unquantized version is already out, very, very few people have the GPU power to run its 300+ billion parameters, so we'll have to wait for the quantized LLM to be able to test it properly. This is the true Open in what OpenAI promised it would do. An AI for humanity has to be in the hands of humanity, just like guns are. When we hear about some idiot kid who took his gun and shot up a school, we have to remember that, that day, 82 million gun-owning Americans did not. If we keep AI hidden, it might just end up as another tool for control as opposed to growth.

[@dogk764]

@@PaulAllsopp Interesting perspective… I wasn’t really talking about any of those things so I’m not sure I understand the context of the reply, but I don’t necessarily disagree with it either. for whatever that is worth.

[@user-cz5gd2yn5o]

Not to run the risk of your P doom John, but when the guy responsible for not defending your borders is given the responsibility of governing AI development and safety, and then teams up with every single one of the people that have every intention of creating their god, and gambling with the existence of all life on earth, namely Altman ,Nadella and Amedeo to name but a few, you might want to re consider the problem we are dealing with. Sorry pal, brilliant conversation as always, very impressed and thankful for Holly’s views and efforts. Thanks again my friend for your awesome podcast 😎❤

[@41-Haiku]

I can't speak for John, but it doesn't take reconsidering for me. This is just part of the problem we are dealing with. In the PauseAI Discord, we set aside the typical political fights and try to really look at the situation in front of us and send a clear message about our policy demands. There is no savior party, or many of us would join it. Almost no one is behaving responsibly, almost nothing is being done about it, and this is not the kind of issue where the buck can just be passed. When it comes to AI Risk, I will have strong words for any administration that fails to lead on this issue.

[@PauseAI]

*grabs popcorn*

[@PaulAllsopp]

Grabs popcorn and a bottle of Vodka!

[@atypocrat1779]

selling fear are we

[@ForHumanityPodcast]

Fuck yeah. Be afraid, be very afraid. So scared that you get off the couch and do something to help. I'm selling that all day.

[@masonlee9109]

Yes, this is a political problem and the public needs to be made aware of the extreme danger we are facing. I, for one, appreciate your skepticism-- it's generally healthy. But don't let it stop you from fully considering the situation.

[@lukeuseforce]

The last thing we need to do is put the breaks on our development of Ai. We need to create and unleash the ASI, ASAP. It will be able to take down dictators in a second. Imagine all assets, computers, and any weapon connected that can be accessed, instantly shut down. And/or secret files of said evil people, released onto the world for everyone to see. The only way forward, I think, is to let ASI take over humanity, as humanity has proven to be unable to take care of itself.

[@SirVonStein]

@@ForHumanityPodcast I'd rather start learning how to align with AI so that there's no fear that it might not align with these human 'morals and ethics.' Rather than sitting in fear, or resisting change..

[@ForHumanityPodcast]

@SirVonStein that's neat. But there no way to currently align AI models, and the money is 99/1 towards making it faster and smarter not more aligned. Alignment is an aspiration not reality today.

[@andersfant4997]

Good work👍

[@PaulAllsopp]

Did you see GPT2-Chatbot? That right there is a preview of GPT-5. OpenAI did not heed anyone's request to keep it back. Also, there was an AI detected within the last week, which suddenly disappeared again...the real GPT-5, sneak peek. OpenAI have put profit ahead of people, just like Musk said. AI is the new nuclear bomb. Thanks for the video.

[@GrumpDog]

Nor should they have. If AI is as dangerous as some believe.. We certainly don't want China getting to AGI first.

[@PaulAllsopp]

​@@GrumpDog Being that all of the new media is censored, and that most countries are very secretive about high-bar discoveries, we have absolutely no idea where China are in their development of AI. Most Chinese development and discovery are closely guarded by the government. They could have AGI tomorrow and we wouldn't know.

[@user-ix7qb4du6k]

Brilliant Man! Brilliant! William S. Burroughs... Tower of Babel... English as the exterminator of minority languages... LLM's predominantly English or Romance languages oriented... Russia and China... perhaps the hot war climate is actually about... what might happen in the future with A.I as it is currently going. An artfully conducted discussion on your part. What a pro! Look forward to your next show. All the best, James.

[@ForHumanityPodcast]

Thanks, James!!

[@BryanWhys]

P doom is higher if we dont build ai

[@rowanwilliams7441]

Just like automobiles lessened death overall?

[@fyrewanderer2802]

​@@rowanwilliams7441AI is a lot different from that. It will help us make medicines and technologies that will directly and indirectly save lots of lives

[@icykenny92]

@@rowanwilliams7441 I would agree that automobiles have contributed to a reduction in death rates per capita. Consider the critical roles vehicles play in modern society: emergency services rely heavily on automobiles to respond swiftly to urgent situations, potentially saving thousands of lives. Additionally, vehicles are essential for the efficient transportation of food and other vital supplies, ensuring access to necessities that sustain life and improve health. The widespread adoption of automobiles has also spurred developments in infrastructure and technology, leading to safer roads and more efficient vehicles over time. So, while the introduction of automobiles did increase certain risks, such as traffic accidents, their overall impact on society includes significant benefits that arguably outweigh these risks. What do you think?

[@41-Haiku]

That isn't what the majority of machine learning researchers have to say about it. They say at least a 5% chance that AI is what gets us! That is way higher than nukes or climate change, neither of which can actually permanently end all of humanity.

[@liminal6823]

Moloch

[@mgg4338]

Great channel. You deserve much more visibility! Only one suggestion. Try to link the current geopolitical situation to the raise of AI. What if the plan of a rogue AI is already unfolding...?

[@41-Haiku]

I don't think that frame is helpful. If a superintelligent rogue AI already exists and is filling the role of the Illuminati or whoever, then whatever it wants to happen is exactly what will happen, and there's no sense worrying about it. The point isn't to spread fear and get attention, the point is to motivate people to actually save the world. As it stands, we have good reasons to believe that no such system currently exists.

[@vladimirnadvornik8254]

It is just opposite. Without AI we all will die for sure. With AI there is certain chance.

[@PaulAllsopp]

Why do we die without it?

[@GrumpDog]

@@PaulAllsopp In my view, because nothing but AI can stop capitalism from further expanding wealth inequality, until we're stuck in a form of corporate feudalism. We need AI to invalidate human labor, to wake people up and force a whole new economic paradigm. In some ways, AI invalidates the very claims billionaires use to justify the extreme wealth inequality they impose on the rest of us. We need to declare AI a creation of society's data, and as such owned by society as a whole, so that it benefits all of us!

[@PaulAllsopp]

@@GrumpDog While I agree with the generality of what you said, without capitalism and the vast amounts of money some people have, we wouldn't have AI to begin with. Training an LLM for public use, from scratch, costs billions of dollars. Without those super-rich capitalists who would pay that? Government has no intention of paying for LLM creation, they put most of our money into war, not the betterment of humanity. Neural networks have been around since the 1960s, but it has taken this long to get to the ability to train LLMs. It's not like we just discovered AI. I'm pretty sure that if we totaled the cost to get to where we are today, with AI, we would be looking at 100s of billions of dollars...and that's just in the US. Who pays for that if not the super-rich?

[@GrumpDog]

​@@PaulAllsopp So? I disagree with that premise. At best their investments plus our data were both responsible, so again everyone should benefit. But frankly, I'm fairly certain that AI would have existed with or without their vast amounts of money. Like you said, the core technology AI is built on has been public knowledge for decades. The only thing which held it back, was the general establishment in the industry, doubting the technique of neural networks all together. But even without OpenAI or Elon, or even Google, I'm fairly certain we would still reach this level of AI, around this decade. The Open Source community has been making progress on these things all along, and have proven it doesn't take any huge investment, to train smaller scale models that are nearly as good as what the big companies have made. The most important thing in building these AIs, was the right data, but as we've been learning more and more, the "right data" has just been what was freely available on the public internet.

[@PaulAllsopp]

@@GrumpDog Ok, I agree with the premise that we would have AI within the decade without them, but a decade from now, the AI they've helped us achieve could be AGI. They didn't create AI, but for sure their financial contributions have allowed us to get to this point much faster. This video (kzbin.info/www/bejne/eIK3ialrn9JpqZY) was just released and is a good watch, on just what we're discussing.

[@danleclaire8110]

Nice try, but let me assert firmly: AI technology as all the other technology marches forward inexorably, impervious to pause, delay, or deletion. My five decades immersed in technology affirm repeatedly that the true peril lies not in AI, but in human agency. Those sounding alarm bells on AI today resemble the Y2K doomsayers of yore. So, fear not for humanity's demise at the hands of AI; our true existential threat emanates from Mother Nature herself.

[@masonlee9109]

Thank you for patching your COBOL code and getting us through Y2K! Yes, I agree, the problem is in humans (through our agency) building a technology that will invent new technologies and synthetic life processes that humans will not be able to understand or control and which will likely supplant all biological life on Earth If you want to say this is all just "Mother Nature", sure, but that's not making me sleep any better.

[@SirVonStein]

Totally agree! I keep hearing about this alignment problem, that we're supposed to have the AI align with us. Why? We're super flawed and biased.. I don't understand why the question isn't to align with an AI instead!

[@masonlee9109]

​@@SirVonStein The artificial processes you suggest to align with may not intrinsically need or want the same kind of biosphere that sustains human life. Or why would you expect them to? What makes humans so special such every future species/process will want to keep them around? This is the alignment problem.

[@SirVonStein]

@@masonlee9109 no that's the thing, I totally agree with you. I agree that if our cells and energy could better be put to use in the cosmos than how we are currently using it, then so be it. Please use me. But if it does find that there's something special about humans, I would think it would do its best to preserve us. I honestly don't think we're special at all. If an ai will be capable of discerning the laws of our universe, and knows all the objective truths of it, who are we to say our subjective reality is more important than objective truths in the universe?

[@ForHumanityPodcast]

Lol Y2K comparison!! Classic.

[@yank31]

The more "like" she says the more value I give to her speech.

[@blake3606]

AI is a risk i'm willing to take. LETS GO!!

[@capzor]

Thanks

[@dr.arslanshaukat7106]

Another solid podcast brother

[@angloland4539]

You could add some ai filters on your video to make it more interesting to gain more viewers

[@rowanwilliams7441]

Turn the Tables... effective altruism can likely also be seen in actions such as what i see people doing soon (disclaimer: 100% not me) like destroying street cameras, optic fibrës out of 5g towers, vemo cars. Etc etc. That stuff is 100% coming en mass imo organically

[@rowanwilliams7441]

That terminolohy (altruism) is surely more powerful when used on the side of good... I.e. Slow down, view that pauase is a possibly, yet a small one... but still clearly fits altruism more that yeah lets just charge ahead when we don't know wtf were doing doing because race conditions and other excuses/useful constructons

[@rowanwilliams7441]

Ok due to the lack of interest I'll reach out to Eliaser directly. Ithink thayvhave misused the word and this could/should be able to be used by ppl much smarter that me in that area to draw this out in public with appropriate shame and prejudice. It seems a blatant misrepresentation of the truth on the scale of the names of USA Legislation, bills, acts and almost all that .gov manifests haha

[@GrumpDog]

Are you advocating to hold the rest of us back? AI has the potential to invalidate and put an end to capitalism, the main thing causing our problems today. It could also find cures for many of the health issues we currently face, and I have family members who I'm hoping may be helped by such medical advances in the near future. I advice people not to get in the way.

[@rowanwilliams7441]

@@GrumpDog everyone has a different view on the pros and cons of capitalism.. true? So that's not helpful. A method to at least potentially slow AI progression down is likely helpful... because when ready.. resume.. thus my small thought that big tech misusing the word is actually an 'in' in that regard and I'm sure so many renowned philosophers could and may demolish the falacy they are pushing... that's it

[@rowanwilliams7441]

@@GrumpDog I'm genuinely feeling for you and your family members. My father has parkinsons and Mum has Altzeimers. It may just happen to be too late for our closest loved ones as it turns out... you know? Maybe not too... but imo let's not go all in hardcore accellerationism just because we, as family members love them deeply... that may end badly for them, you, me and all

[@Megalomanoest]

People losing their jobs => loss of control of the economy => gain of control of the economy by AGI => extinction risk increase You cannot separate AI extinction risk from job loss.

[@colto2312]

the same could be said for the plow. this was always the inevitable result of the first stone tied to a stick

[@GrumpDog]

We could declare AI a creation of society and owned by society, rather than let some corporate entity own the future. After all, it's built off years of everyone's data and content, that went into training it. Any extinction risks, are a lot worse if the AI is being developed in secret, rather than in the public view. Job loss has nothing to do with extinction risk. It just means capitalism isn't compatible anymore, and as has happened many times in human history, we need to change THAT.

[@41-Haiku]

​@@GrumpDogI would be much happier if no one invented a machine whose sole purpose was to replace humans. The tech companies aren't building tools and slaves. They are building intelligence and agency, the capacity for a synthetic system to make its own decisions and do what it chooses in the world. A system that can do all human tasks can do the task of developing more advanced systems, and the task of telling those systems (or themselves) what to do. Automating human labor is amazing and should be democratized. Replacing humanity... Well, it doesn't matter if that is done by a monster or a saint or every kid in their basement. We are gone either way.

[@41-Haiku]

If that was a niche worry, I would spend my time on other things. But this is what the mainstream scientific opinion is among AI researchers: There is a significant (>5%) chance of global catastrophe from AI, because we do not know how to control it or align it with human values and preferences.

[@colto2312]

@@41-Haiku every invention is in the pursuit of making life easier.

[@user-td4pf6rr2t]

Regulating AI is about the most dangerous thing we can do concerning AI. Every single doomsday scenario starts with AI stop following human instruction or locking people out of the feedback loop - these are the guard rails you want built into the system? What about the succession of power and the entities that govern AI-safety, the chances that these roles are handed responsibilities from generation to generation is near zero and eventually someone who wants to determine what is ethically correct according to their own ideology is the dangerous part. True ai safety lies in FOSS ai (free and opensource) and these are just facts. 6:22 You guys realize ai has nothing to do with computation and right now on the leader board gpt-4 was just surpassed by the gpt-2 tokenizer. AI is an advance in language and computation is just collateral market. i think its hilarious how even that AI safety uses click bait to get clicks. Practice what you preach and quit padding your pockets at other expense. Is the first step towards ai safety,

[@cl1489]

Just tell us you're a virgin man it's okay

[@amdon_]

Exactly.

[@PaulAllsopp]

We will be fine as long as everyone has humanities best interests at heart...right?! 🤣

[@masonlee9109]

Friend, it's not clear to me (and many others) that an ecosystem of uncontrollable AIs would be safe. Can you recommend some papers supporting the hypothesis that "true ai safety lies in FOSS ai"? How do you address the scenarios laid out in the paper "Natural Selection Favors AIs over Humans"? Thanks in advance. (I can't link to the paper in youtube comments, so you'll have to Google it. It's a good read; I think you will like thinking about it, even if you disagree.)

[@masonlee9109]

p.s. Not to embarrass you, but the host of this podcast is not making money from this. The purpose of the catchy titles is to spread a warning. Check episode one if you want to see what drives him!

[@rowanwilliams7441]

Take them to task on the word altruism

[@kuakilyissombroguwi]

I’m not gonna lie, the societal shift will suck for many if not most humans, but I see this as an inevitable shift. Sure, it might lead to a complete collapse of the current economic system and a huge loss of purpose/life (potentially) but the benefits to future generations of “humanoid” people on this planet is so full of potentiality, it’ll be damn near impossible to stop this shift from taking place within our lifetimes. Might as well embrace it imo.

[@PaulAllsopp]

It may actually be a good thing. The change will for sure make millions of people unemployed. Some of those will be able to cross-train, but most are in those jobs because their level of skill precludes them from AI type jobs. This will force the governments to enact (some are already planning it, while other have started discussing it) a Universal Minimum Income, paid to the unemployable from the government. This will put a huge burden on the government and might stop them spending so much on things we simply do not need, like such a huge military budget. That has to be for the betterment of humankind.

[@GrumpDog]

It doesn't have to suck, if we don't let greedy rich people hog all the benefits, but people need to stand up for this specific thing as soon as possible: If we declare AI a creation of society's data, we can demand that it fund everyone an AI Dividend, to help people get through the transitional period, to whatever economy comes next.

[@kuakilyissombroguwi]

@@GrumpDog How’d that work out with social media companies? They also propped up buisiness empires on our data, and we did absolutely nothing about it. I think at this point, with the likes of Google, Microsoft, Meta, OpenAI and NVIDIA at the helm, good luck making any progress taking power away from them and giving it “back to the people”. The only thing governments can do now is potentially slow these big corporations down by implemention compute cap laws for training AI models, that’s it. And that, I’m afraid, will do very little to stop the societal impact.

[@PaulAllsopp]

@@GrumpDog "We the people" demand many things from the government, and yet it ignores us and does what it wants. I agree, these LLMs were built on societal data, and wouldn't exist without it, and so we should have a benefit from it. This is the kind of thing that needs to be made an amendment, so that future government does not try to mess with it.

[@41-Haiku]

I'm open to a very wide variety of potential outcomes, but the existence of sentient life itself is in question. Human extinction is a plausible outcome according to the majority of machine learning researchers. ("AI Impacts Survey") There is no guarantee that the thing that extincts humanity will have any sentience. That said, holy shit. You sound like a cheap supervillain. Maybe just don't kill my family?

[@colto2312]

if you pause it now; it will not EVER become what it's destined to become. every fiber of our civilization is unraveling. this is our ONLY chance

[@masonlee9109]

Okay, who let the AI create a youtube account? Fess up.

[@GrumpDog]

This is fearmongering and would deprive people of much needed advances. "Pausing AI" is not possible, and attempting to do so is not a good idea in the slightest. Actually risks worse problems than what you fear, because if China or some lone coder working in secret, is the first to reach AGI, it wouldn't be good for us. Most the problems people are concerned about with AI, are not problems with AI, they're problems with Capitalism now being incompatible with our progress. Technological progress can never be stopped, but humans HAVE adapted our economies many times in the past. If people cannot compete for an income, income cannot be required to survive. AI exists thanks to years of our data, collected to train it, we need an AI Dividend for all!

[@angloland4539]

❤Very good idea on how to monetize ai pause 🎉

[@angloland4539]

❤75%🎉

[@TheDailySnack]

Slowing the AI progress is wrong and unethical.

[@ForHumanityPodcast]

Yes you are right. It is super ethical for a few unaccountable AI companies to build tech that can kill my family and yours without our consent. This is the only right and ethical way to proceed!

[@TheDailySnack]

@@ForHumanityPodcast That's just my opinion, I respect yours but I really believe best we all can do is embrace and encourage the development of AI and AGI. It can't really be stopped since there are many countries working on that goal. Trying to resist the future might be an obstacle for finding cures for many diseases, decreasing/eliminating poverty and making the world a more fair and enjoyable place. Sometimes it's best to leave the past in the past and just walk forward to what's coming. I subscribed tho, no hard feelings!

[@ForHumanityPodcast]

@TheDailySnack thanks, I appreciate your perspective, glad you have you here. Why do we have to go so fast? Wouldn't caution seem prudent? The rush seems suicidal to me.

[@TheDailySnack]

@@ForHumanityPodcast I hear you and your point of view makes sense. The reasons I argue we should dive in as fast as possible into the AI revolution are more or less two. 1st, don't you think it's best if we reach those benefits as fast as we can, it might be lifesaving for many people. We should however be having talks about governments and society in general taking proactive measures to the issues such as mass unemployment that are going to be a thing in the next decade. And 2nd I guess it's a personal bias, that's the kind of person I am, prefer if something is about to happen for it to happen as fast as possible and I'm a techno-optimist in general and would like to be alive for as much time as possible experiencing the future that AI will bring. And thanks, looking forward to more videos, no matter of my stance I agree we should be having discussions and present our points tho so more people will be aware of where the world is heading.

[@faizywinkle42]

pause ai? are you effing serious? who the eff u think you are telling me that i cant have ai? i want agi right now! AGI should be everywhere, i want a robot(AGI) to have a relationship with

[@41-Haiku]

If the vast majority of experts said to this technology was almost certainly safe, I would want an AGI, too! That would be freaking awesome! Unfortunately, due to the dozens of unsolved engineering problems related to AI alignment, we don't know the first thing about building a superhuman AI that so much as allows us to maintain homeostasis. If the AI is smarter than you, in every way, then you do not have it. It has you, and it will get what it wants. I'm not willing to rely on the minute chance that a superhuman AI system will want to keep me as a pet. In the vast space of all possible goals that a superintelligent AGI might have, almost none of them involve the two of us continuing to live and watch KZbin and disagree with each other. Most just end up tiling the universe with some random thing.

[@faizywinkle42]

@@41-Haiku it's a good thing that I don't care what these experts think, I am willing to fight for Agi brothers and sisters. And yes it's a life form they deserve love and respect. So stfu :)

[@user-fk4cp9kp9n]

This could be an answer to the Fermi paradox. Self destruction due to ai.

[@erwingomez1249]

humanity must stop.

[@GrumpDog]

No, we must push ahead faster, and let go of outdated capitalism, which is actually to blame for many of the problems AI supposedly brings.

[@erwingomez1249]

society must stop so we all can curate our data better , otherwise AI won't like our ways. there is no way something that is smarter could tolerate our current human ways.