The AGI Lawsuit
Added 2024-03-03 21:58:16 +0000 UTCWhat are just the most interesting details from the Musk-Altman Lawsuit? Can Gemini 1.5 help me sort through the morass of relevant tweets? I want to give you the history of the battle over the definition of those three key words - artificial general intelligence - and what it means for us all.
Lawsuit: https://www.courthousenews.com/wp-content/uploads/2024/02/musk-v-altman-openai-complaint-sf.pdf
NYT Sparks: https://www.nytimes.com/2024/03/02/technology/elon-musk-openai-lawsuit-microsoft-research.html
Altman Interaction: https://twitter.com/sama/status/1599671496636780546?lang=en
Grok Flaws: https://www.washingtonpost.com/technology/2023/12/23/grok-ai-elon-musk-x-woke-bias/
Sutskever Admission: https://www.theverge.com/2023/3/15/23640180/openai-gpt-4-launch-closed-research-ilya-sutskever-interview
Musk Interesting xAI Found Date: https://www.tbsnews.net/tech/elon-musk-forms-xai-artificial-intelligence-company-616798
OpenAI Charter: https://pbs.twimg.com/media/F_fs3TWW4AAn08b?format=jpg&name=medium
Explained Prediction: https://www.youtube.com/watch?v=Auu8tGaTaNQ&t=4s
Musk Leaves OpenAI: https://www.wsj.com/tech/ai/chatgpt-sam-altman-artificial-intelligence-openai-b0e1c8c9
‘Voluntary Moratoria’: https://futureoflife.org/open-letter/pause-giant-ai-experiments/
Nadella Interview: https://www.wired.com/story/microsofts-satya-nadella-is-betting-everything-on-ai/
Altman Only In Group: https://twitter.com/sama/status/1599112749833125888?lang=en
Worldcoin Inequality: https://twitter.com/worldcoin/status/1544071558179594243
Musk Tax Point: https://twitter.com/Teslaconomics/status/1763953141819818387
DeepMind AGI Definition: https://arxiv.org/pdf/2311.02462.pdf
Comments
“When would the US government take over the company that launched real AGI/ superintelligence and how would they do it?". Brilliant. So many scenarios/ possibilities/ parallels spring to mind. Would love to see a video about that. Or the first feature film from AI Explained? ; )
Martin Percy
2024-04-03 17:44:57 +0000 UTCI'm not sure if Mr "If we go extinct it would be OK because its just the next step of evolution" Larry Page of Google is actually trustworthy with this tool.
Carlos Andrés Campo González
2024-03-15 17:05:45 +0000 UTCNoah Smith always good, if not exactly what I was looking for: https://www.noahpinion.blog/p/why-trying-to-shape-ai-innovation
Mike Pemberton
2024-03-07 17:52:04 +0000 UTCSome evidence provided by OpenAI. https://openai.com/blog/openai-elon-musk
Carlos Baraza
2024-03-06 03:25:47 +0000 UTCAside from the debate around defining AGI, there's surely great scope for sub-general artificial intelligences to wreck havoc with labour markets. I don't think Elon's lawsuit will help much with that, whatever the outcome. I've seen lots of speculation about what a post-AGI utopia/dystopia might be like, but nothing on how the transition can be managed. It's all very well folk saying "oh, we'll just give everyone universal basic income while the AI does the work" but if during the transitional phase all the value from AI labour goes to half a dozen companies in silicon valley, how does the UBI get paid for? Big companies are very effective at dodging corporation tax. The basic economic view that value is derived from land, labour and capital is an interesting starting point. If labour and capital collapse into a single entity (labour = AI, capital = data centres) owned by Google, what remains is land, i.e. natural resources. Has anyone got links for good quality writing on the topic of transition to the AI economy?
Mike Pemberton
2024-03-05 17:01:11 +0000 UTCI'm slightly, but not entirely sceptical whether Elon's motives are as altruistic as he would have us believe. Certainly he's conflicted, owning a rival AI company.
Mike Pemberton
2024-03-05 16:16:49 +0000 UTCThanks for another great video. I almost wish this particular one wasn't behind the paywall, since it is such an important topic in general. As other commenters said, hopefully this will bring more clarity on the "AGI" term itself. It is a good point that these clauses are way too amorphous. Like even putting aside the lawsuit, if OpenAI suddenly claimed AGI on a new model, there is a decent chance that MS would sue them to not lose the IP. Then you are in the courts anyway.
Shawn Fumo
2024-03-05 15:12:57 +0000 UTCI don't know... I'm definitely not a fan and I agree he seems narcissistic, but he's also been pretty consistent on being worried about AI for many many years. Like almost everyone involved in this space (OpenAI, Anthropic, etc), there's this element of "it is dangerous, but I'll be the one to do it right". Like him yelling at the intern, I'm sure in his mind of course it'd be safe at Tesla since he's in control of Tesla.
Shawn Fumo
2024-03-05 15:07:35 +0000 UTCI think it will be difficult to expect proof of consciousness from AGI. There is no clear definition, and there is no way to prove any other human is conscious. It is purely a subjective phenomenon. I think some independent agency will be a more workable phenomenon.
Arek Stryjski
2024-03-05 13:57:54 +0000 UTCSome would and had argued that you need consciousness and awareness to be called AGI. And there doesn’t seem to be a consensus on the definition. It makes me reflect that the root cause of so many of our disagreements is slightly different interpretations of common terminologies or definitions
Kevin Li
2024-03-04 21:21:30 +0000 UTCRegardless of the position of both parties and their interests I do think it's great that we are going to see a long and public discussion about what is AGI and how far away it possibly is. I see many people currently using the "AGI" term to mean very different things or in quite vague ways. Maybe this discussion can bring more clarity.
Tomas Dulka
2024-03-04 18:07:24 +0000 UTCThe claim that GPT-4 is AGI is quite funny given how many intellectual tasks this very useful tool cannot do at all. As of how to define an average person level AGI - perhaps we could define it by pointing to some professions with diverse tasks? If AI can do all of them then it can also do the hundreds of others that are hard to enumerate (assuming it wasn't specifically trained for it). How about a travel guide who must organize a trip and lead a group of tourists? It requires detailed planning, updating the plans as well as some people's management. It doesn't encompass fine motor skills nor constructing scientific hypothesis but that could be added to this travel-guide-AGI definition. WDYT?
Jan Matusiewicz
2024-03-04 17:48:28 +0000 UTCPosted this morning and out of date by the afternoon... Thats' AI folks
Rakesh Murria
2024-03-04 16:45:07 +0000 UTCVery interesting, thank you for sharing.
Jean-Emmanuel COMBE
2024-03-04 13:53:14 +0000 UTCAlso, I'm quite concerned about Gemini, relevant to this case - I believe. https://x.com/AndrewCurran_/status/1764170247597404416?s=20
Patrick & Sylvia Trentelman
2024-03-04 13:50:30 +0000 UTCThere is the Option article in NY Times suggesting Musk is a number 2 in MAGA. If true this could suggest he just started new chapter in Cultural Wars and it has nothing to do with technology, just politics. Why Elon Musk Is the Second Most Important Person in MAGA https://www.nytimes.com/2024/03/03/opinion/musk-x-maga-trump.html?unlocked_article_code=1.aE0.Hp-T.FlAs0iVjQ8qH&smid=nytcore-android-share
Arek Stryjski
2024-03-04 13:45:03 +0000 UTCI found this 🧵of a tech laywer disecting the suit a good read: https://x.com/CeciliaZin/status/1763849318396752151?s=20
Patrick & Sylvia Trentelman
2024-03-04 13:40:35 +0000 UTC"transmogrification" I love the ref haha
Wassym Trabelsi
2024-03-04 11:53:27 +0000 UTCI have just one thought: 'I want the video on the Gemini 1.5 Pro!'
Jean-Emmanuel COMBE
2024-03-04 10:49:20 +0000 UTCThis is a properly great video Philip. Exactly what I was looking for to get my head around what he was trying to do. Smart, sly etc. Interesting how the Open AI response was about GPT4. This video is also just a massive side boost for showing that Gemini 1.5 seems to be by far the best model (at least some) people can use right now. Do you agree on that?
Rakesh Murria
2024-03-04 09:05:57 +0000 UTCI see a legal block to the advancement of AI (defined loosely) in democracies with strong independent legal foundations - although this is based on the assumption of copyright enforcement. If AI systems are devised that can perform well on synthetic data then maybe this won't be a blocker. Until then, I speculate that western democracies will start to lose ground to kleptocracies. It will be an interesting challenge to the power of capitalism to overcome any legal constraints in the pursuit of shareholder value.
Niall Riddell
2024-03-04 08:16:46 +0000 UTCGreat video 🙌🙏
Brian Crabtree
2024-03-04 06:30:13 +0000 UTCGemini 1.5’s huge context window reminds me a lot of meta learning. You can just feed it completely new information and it learns. That feels like a pretty general intelligence to me. Awesome video!
Bryson Tang
2024-03-04 04:29:37 +0000 UTCIt's so obvious that Musk only tries to slow OpenAI down. Musk is a narcissist who cares only for himself and certainly not for humanity, lol
Phillip Lakaschus
2024-03-04 03:25:36 +0000 UTCI love you man I hope to meet you one day in the post-singularity Archeo-futurist Miyazaki Patchwork. May we all hollow out asteroids the size of New Jersey, terraform the inside, and retrofit the outside as interstellar travel capable Starships.
r
2024-03-04 03:20:38 +0000 UTCAnother homerun video
r
2024-03-04 03:01:14 +0000 UTCGreat video. But I wouldn't be so sure that AGI would change the fabric of capitalism. I could totally imagine that the companies that control AGIs make enormous gains. Shareholders are happy but those who aren't rentiers nor have a job unaffected by AGI have to live on unemployment benefit. And while they have what to eat and where to live - they don't have money for most of the wonders of the age of AGI. Human needs are insatiable and new ways to spend for the rich minority (like cosmic tourism) would surely be invented.
Jan Matusiewicz
2024-03-04 02:36:10 +0000 UTCI’m not a lawyer, but I wonder if restitution here would amount to EM being given the proportion of the company that he funded i.e. if he gave them 30% of their funding then he would get 30% of the company at its current valuation. That could be a very large amount of money given that OpenAI is now valued in the billions.
James Maclaurin
2024-03-04 02:32:26 +0000 UTCGreat video - Try feeding in the Roon/@tszzl tweet as well - I bet there are more that are relevant to the lawsuit.
Alexander Fullam
2024-03-04 01:16:33 +0000 UTCAt the very least, it will bring the word "AGI" more into the mainstream. It's a concept normal people should be starting to get familiar with.
Anouar Mansour
2024-03-04 00:50:25 +0000 UTCVery spicy take, I love fired up Philip
David Shapiro
2024-03-04 00:23:45 +0000 UTCOpen source is the way
David Shapiro
2024-03-03 23:59:32 +0000 UTCExcellent analysis, as usual. I'm not a lawyer but there is the doctrine of Standing, also known as locus standi. It states that a party can only bring a suit if they have suffered a concrete and particularized injury that is actual or imminent, not hypothetical or abstract. The purpose is to ensure that courts are only addressing actual controversies that affect the parties involved rather than hypothetical disputes or broad societal grievances. A non-profit having a wholly-owned, publicly-traded, for-profit subsidiary is not unknown. E.g, a non-profit Catholic hospital could own a publicly-traded, for-profit health insurance company. This arrangement exists, and OpenAI is not the first non-profit to use it.
Joe Marler
2024-03-03 23:14:39 +0000 UTC
