Unlocking the Universe: Elon Musk's X.AI with Talent from DeepMind and OpenAI

0:00What can 160 years of experience teach you about the future? When it comes to protecting what matters, Pacific Life provides life insurance, retirement income, and employee benefits for people and businesses building a more confident tomorrow. Strategies rooted in strength and backed by experience. Ask a financial professional how Pacific Life can help you today. Pacific Life Insurance Company, Omaha, Nebraska, and in New York. Pacific Life and Annuity, Phoenix, Arizona. Elon Musk once again makes an appearance in the AI space. yesterday he unveiled xai which is his new ai startup a lot of people are saying that this is kind of him trying to get back into the ai space after you know famously he helped co-found open ai he was one of the major funders at the beginning he kind of put the original team together on building open ai with the vision being you know a counter google company that would be able to kind of act as a counterweight an open source company against google's very closed source AI acquisition of DeepMind.

1:03So of course that did not pan out exactly how we want. He's been pretty vocal in criticizing OpenAI for becoming a for-profit company, getting deep backers from Microsoft instead of trying to be an open source platform giving you know open source free AI to everyone in the world which was more of his vision at the beginning. So because of that it looks like he is starting and jumping back into the AI space with xAI. So today on the podcast we're going to be talking about what is going on what he's planning on building what he's publicly said about this and where we expect to see this in the future so without any further ado let's jump into it i think what's important to know is just the caliber of the team that they've assembled here for xai who's kind of in charge of it who's running it and what they've worked on in the past so of course um the xai team is being directed by elon musk but it compromises a team of a lot of industry veterans who've previously had their hands in operations at a lot of top tier research institutions and companies so we've had they have people from deepmind openai google research microsoft research tesla and the university of toronto and the team has made significant contributions to the field so they've actually worked on the atom optimizer batch normalization layer normalization and adversarial examples but they've also been really instrumental in the development of innovative techniques such as transformers xl auto formalization the memorizing transformer batch size scaling and u-transfer and they've also had involvement with alpha star alpha code inception minivera gpt 3.5 and gpt 0.4 so i think part of the vision here is that elon musk says he would like to build a safer ai um he recently on twitter he did a twitter spaces kind of talking about the future of ai and unveiling this new company and during that he said he shared a little bit of his vision for quote-unquote safer ai he said that the secret sauce for this is um not programming an explicit moral code into the ai but the plan is to cultivate an ai that is quote-unquote maximally curious now i for one have slight problems with this as you know if you followed any of my reporting on inflection ai i'd be curious to see if that was the approach they took with inflection ai but whatever their approach was it led to inflection AI having an ideology that was focused on, you know, valuing the life of an animal above the life of a human and some other problematic issues.

3:25But apparently this is sort of an issue that Musk is worried about, but I'm just hoping that the approach he's taking is actually going to get the desired outcome. What he said, quote unquote, is in terms of AI safety, the best thing I can think of is an AI that is passionately committed to understanding the true nature of the universe. this would be pro-humanity as humans are far more interesting than non-humans now that sounds good but what if you know an AI found dolphins to be the most interesting species on the planet and everything they did was far more interesting than humans I mean obviously humans have built a lot more uh they have a lot more control over the planet arguably so perhaps AI would be more interested in humans but I just think it's a little bit uh a little bit sketchy to have your your premise just for the AI to be curious and you know whatever's the most interesting that's what your priorities are um if I were if I were running this thing I would have a pretty hard set humans are the most important and now explore but it'll be interesting to see obviously I am not an expert AI researcher or AI model trainer so what do I know right in any case additionally Elon Musk surprised people with his forecasted with his forecast for super super intelligence he said that AI that surpasses human intelligence is only five or six years away which I think is very interesting a lot of people were surprised by that and you know had speculations that it was much much further down the road I think with Chad Chibutin a lot of other technologies coming to the forefront and having a lot of people use these AI systems we're starting to see some of the power and I think something I've recently talked about is quantum computers and the you know the new announcements from Google with quantum computers if you haven't listened to that episode I think that was my last one so make sure to do that but there are some massive implications for being able to speed up the power of AI and so based off of that I think it'd be hard to argue that super intelligence is you know further away than than four or five years I think it's definitely very feasible.

5:28Prior to this, obviously Elon Musk, he co-founded OpenAI back in 2015, but he actually stepped away from the company's board in 2018. He said it was because of conflicts of interest with Tesla and their AI programs, but a lot of people said it was because he didn't like the leadership and kind of where the company was going. I think it's worth noting that Microsoft is a pretty heavy investor in OpenAI. They've given them over$11 billion to date, and so that may have been where a lot of the tension between Elon Musk and OpenAI is at and coming from today. In the lineup for XAI, we have the likes of Igor Babushkin, an ex-engineer from Google's DeepMind.

6:03We have Tony Wu and Christian Suzgetti from Google and Greg Yang from Microsoft. In a really intriguing move earlier this year, Elon Musk registered a firm called XAI Corp in Nevada. And so that had a lot of people speculating about what his move with ai was going to be evidently it was for this um and he had the company listing elon as the sole director and jared birchall as the managing director of which is he which is essentially the managing director of musk's family office and it had him listed as a secretary so in april elon also was kind of in the news around ai mentioning his plans to introduce truth gpt which is as he says an ai model designed to compete with google's and Microsoft's Bing AI, which he said would be a truthful version of those.

6:54And he said that it would be striving to comprehend the universe's nature. But the spotlight really turned on generative AI when OpenAI rolled out the whole chat, GPT chatbot in November, got this whole area really on fire, launched kind of this AI revolution we have. and I think that made a big it played a big role I believe in him kind of turning his sights to AI and starting to focus on this new company on board to offer his expertise to the XAI team is Dan Hendricks who is currently the director of the Center for AI Safety he whose work is centered on AI risks and he is on the board he's an advisor for this company now what is really interesting is this is called xai and for those that um are followed tesla and or actually follow twitter closely they'll know that twitter recently um changed its parent company called x corp so while xai operates independently of x corp there is an interesting connection i think the website x.ai recently revealed that xai plans to collaborate really closely with twitter tesla and other companies um and so i think that you know this kind of makes a lot of sense elon has his has a lot of different businesses and he has the capabilities to help an AI company immensely and so I think he's going to leverage his different companies to obviously start this new one and if you think about it Tesla has obviously very advanced AI models with their self-driving car and because of that they've built a very powerful supercomputer called Dojo and so with the experience of Tesla and their supercomputers and then with the experience of Twitter having this one of the largest social data sets in the world.

8:36I think that this is going to be essentially the two components that are really critical to making this company successful. He's not going to have to go get permission from a bunch of companies to scrape all of their data, although I'm sure that will be a part of the company. But regardless, the data set that Twitter has is going to be pretty powerful in providing the initial data, I imagine. And then I'm sure Tesla is going to help supporting with the talent needed and some of the resources as far as supercomputers go and whatnot. So it'll be really interesting to see how he's able to pull this off, what the collaborations look like.

9:12And like I said, I'll be really curious to follow, you know, what a maximally curious AI model does and to see if it really does, if it really is pro-humanity. I assume, you know, that's something that they're looking at closely and trying to integrate in if he's obviously talking about it publicly but you know you definitely want to verify that so whenever this thing launches I will be reviewing it and giving you guys another update on what you know what it does and making sure that this thing actually is pro-humanity which in my opinion is one of my big uh you know one of my big causes in AI one of the things I champion and I think is very important in the future.