OpenAI Publishes Emails Showing Elon Pushed For-Profit Conversion

Today on the AI breakdown, OpenAI fires back at Elon over his lawsuit saying that he knew about the for profit plans all along and in fact was one of the main pushers of them. Before that, on the brief, a new open letter asking the big AI labs to make it safe for independent researchers to check their systems. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to breakdown.network for more information about our Discord, our newsletter and our YouTube channel. Welcome back to the AI Breakdown Brief. All the AI headline news you need in a round. 55 minutes we have ourselves another open letter in the AI field. I'm joking of course, because last year we had a slew of open letters, most notably the six month pause letter, but this one is, while perhaps coming from some similar places, a little bit different. The Washington Post reports top AI researchers say OpenAI meta and more Hinder Independent evaluations and basically what this group of 200 or so people are asking for is for generative AI labs to allow independent researchers and investigators to check out their systems and as part of a broader safety mechanism. The open letter is called a Safe harbor for Independent AI Evaluation. The signatories say that they agree on three main things. First, that quote, independent evaluation is necessary for public awareness, transparency and accountability of high impact generative AI systems. Number two, that currently AI company policies can chill independent evaluation and that three AI companies should provide basic protections and more equitable access for good faith AI safety and trustworthiness research. Basically what this group is saying is that a key part of the AI safety apparatus across companies, not just within any one lab, is to allow people who aren't affiliated with that company and who don't fear reprisal from that company to dig into their systems with potential risks and threats in mind. They write. Disempowering independent researchers is not in AI company's own interests. To help protect users, we encourage AI companies to provide two levels of protection to research. First, a legal safe harbor would indemnify good faith independent AI safety, security and trustworthiness research provided it is conducted in accordance with well established vulnerability disclosure rules. Second, companies should commit to more equitable access by using independent reviewers to moderate researchers evaluation applications, which would protect rule abiding safety research from counterproductive account suspensions and mitigate the concern of companies selecting their own evaluators. While these basic commitments will not solve every issue surrounding responsible AI today, it is an important first step on the long roads towards building and evaluating AI in the public interest. So this is interesting because this is not a request for full openness, it's a request for effectively an openness carve out even within closed systems. The list of letter signers, which they currently have at more than 250, includes some prominent names from the AI space. Now, when it comes to the list of signatories, the vast majority of the 250 plus that they now list come from the world of academia. One notable exception to that is Klem, the CEO and co founder at Hugging Face, which makes sense given that company's position on OpenAI systems. But given how many of these folks are outside of the actual AI business space, it'll be interesting to see if it gets any sort of traction among the labs. I certainly think that potentially it could be a PR win for some of them. Holding aside questions of, you know, whether it's the right thing to do or not, Next up in this brief we turn our attention to the ongoing AI arms race between the US And China, where of course the dominant force right now is US export restrictions around semiconductors and AI related chips. The big US Companies, in response to prohibitions, have tried to find AI chips that they could tailor to that market being powerful enough to appeal to Chinese buyers while still being lower than the bar set by the US Commerce Department. It appears that AMD's slightly lower powered China focused chip is not getting the Commerce Department seal of approval. Bloomberg writes that U.S. officials have told AMD that the AI chip it made for the Chinese market is still too powerful to sell without a license. The market did not like this and AMD shares were down over 2% as the markets opened. Now one of the interesting market dynamics here is to what extent these US Companies are just forced to surrender the market because the offerings that would get the US Green light are simply less powerful at worse cost than homegrown alternatives. Indeed, Reuters reports today China to step up Quantum computing, AI and Tech Self sufficiency drive this is from a government work report that is clearly using these geopolitical issues as a rallying cry domestically. But if these rules are creating something of a headwind for chip makers, the potential for damage doesn't appear to have markets all that freaked out. One noted analyst Ming Chi Kuo wrote on Twitter that they believe that, quote, if Apple fails to launch generative AI services this year that are better than market expectations, Nvidia's market value will most likely surpass Apple. By way of comparison, Apple's market cap right now is 2.78 trillion as compared to Nvidia's 2.06 trillion over the last year, Apple stock has gone up about 13% while Nvidia's stock is up 260%. Apple is in an interesting position. The vast, vast majority of their revenue still comes from the iPhone. But that means that if iPhone sales run into any troubles, that can be a huge challenge. For example, while coming into the year, analysts expected Apple to ship between 220 and 225 million iPhones. That number is now looking closer to 200. IPhone sales to China in the first six weeks of the year were down 24%. Of course, it does appear that Apple has groked that AI is going to be key to these efforts and perhaps key to future iPhone sales, given that they abandoned their multibillion dollar multi year car project, Project Titan, and shifted many of those engineering resources over to now the other company with Apple and Nvidia at the top of the market cap heap is Microsoft, who were also in the news today surrounding their latest motion to dismiss the New York Times lawsuit against them and partner OpenAI. The big reference point in this motion to dismiss is the vcr, the motion reads. Despite the Times contentions, copyright law is no more of an obstacle to large language models than it was to the VCR or the player piano, copy machine, personal computer, Internet or search engine. They also challenged the lawsuit on the basis that the New York Times never gave an example of direct copyright infringement by a copilot user. Microsoft's motion, says the Times, contributory infringement theory thus fails on the very same basis the challenge to the VCR failed four decades ago. It improperly seeks to impose liability based solely on the design or distribution of a product capable of substantial lawful use. For now, this lawsuit continues to be a very important one when it comes to the future of the generative AI space. Lastly today the Wall Street Journal is reporting that Perplexity is raising another round, this one that would value it at over a billion dollars, making it the latest AI unicorn. If this round is completed, it will come just a couple months after perplexity raised $74 million at a valuation of 520 million. Not confirmed yet, but something certainly to watch for. Now though, that is going to do it for today's AI breakdown brief. Up next, the main AI breakdown late last night OpenAI tweeted We are dedicated to the OpenAI mission and have pursued it every step of the way. We're sharing some facts about our relationship with Elon and we intend to move to dismiss all of his claims. That piece, which tagged in Elon Musk led to a blog post on OpenAI's site called OpenAI and Elon Musk. We're going to get into lots of the discussion around this, but let's read the source material in some detail. OpenAI writes, the mission of OpenAI is to ensure AGI benefits all of humanity, which means both building safe and beneficial AGI and helping create broadly distributed benefits. We are now sharing what we've learned about achieving our mission and some facts about our relationship with Elon. We intend to move to dismiss all of Elon's claims. First bullet we realized building AGI will require far more resources than we'd initially imagined. When starting OpenAI in late 2015, Greg and Sam had initially placed to raise 100 million. Elon said in an email, we need to go with a much bigger number than 100 million. To avoid sounding hopeless, I think we should say we are starting with a $1 billion funding commitment. I will cover whatever anyone else doesn't provide, end quote. We spent a lot of time trying to envision a plausible path to AGI. In early 2017, we came to the realization that building AGI will require vast quantities of computer. We began calculating how much compute an AGI might plausibly require. We all understood we were going to need a lot more capital to succeed at our mission. Billions of dollars per year, which is far more than any of us, especially Elon, thought we'd be able to raise. As the nonprofit Next section, we and Elon recognized a for profit entity would be necessary to acquire those resources. In late 2017, we and Elon decided the next step for the mission was to create a for profit entity. Elon wanted majority equity, initial board control, and to be CEO. In the middle of these discussions, he withheld funding. Reid Hoffman bridged the gap to cover salaries and operations. We couldn't agree to terms on a for profit with Elon because we felt it was against the mission for any individual to have absolute control over OpenAI. He then suggested instead merging OpenAI into Tesla. In early February 2018, Elon forwarded us an email suggesting that OpenAI should quote attach to Tesla as its cash cow, commenting that it was, quote, exactly right. Tesla is the only path that could even hope to hold a candle to Google. Even then, the probability of being a counterweight to Google is small. It it just isn't zero. Elon soon chose to leave OpenAI, saying that our probability of success was zero and that he planned to build an AGI competitor within Tesla. When he left in late February 2018 he told our team he was supportive of us finding our own path to raising billions of dollars. In December 2018, Elon sent us an email saying even raising several hundred million won't be enough. This needs billions per year immediately or forget it. Last bullet we advance our mission by building widely available beneficial tools. We we provide broad access to today's most powerful AI, including a free version that hundreds of millions of people use every day. For example, Albania is using OpenAI's tools to accelerate its EU accession by as much as 5.5 years. Digital Green is helping boost farmer income in Kenya and India by dropping the cost of agricultural extension services 100x by building on OpenAI lifespan the largest healthcare provider in Rhode island uses GPT4 to simplify its surgical consent forms from a college reading level to a sixth grade one. Iceland is using GPT4 to preserve the Icelandic language. Elon understood the mission did not imply open sourcing AGI. As Ilia told Elon, as we get closer to building AI, it will make sense to start being less open. The open in OpenAI means everyone should benefit from the fruits of AI after it's built, but it's totally okay to not share the science. To which Elon replied, yup, they close. We're sad that it's come to this with someone whom we've deeply admired, someone who has inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress towards OpenAI's mission. Without him. We are focused on advancing our mission and have a long way to go as we continue to make our tools better and better. We are excited to deploy these systems so they empower every individual. So next we're going to look at the emails that are attached to this. But first let's try to sum up. Basically what OpenAI is saying is that Elon is now acting as though he didn't know that this was the plan all along, but this was the plan all along. In fact, OpenAI is going even farther and saying that not only did Elon agree with the plan to create this for profit entity, he was one of the main pushers of it, suggesting that the only way to get sufficient capital for this battle was something like this. So the real reason they're suggesting for Elon's suit is not because of this betrayal of mission, but because OpenAI didn't do it Elon's way. They didn't go along with his vision to combine it with Tesla. Now of course all of this is made much more real by the inclusion of some original emails. We already heard an excerpt from that email from Sunday, November 22, 2015 where Elon said we need to go with a much bigger number than 100 million to avoid sounding hopeless relative to what Google or Facebook are spending. But let's read in a little bit more detail some parts of his email from January 31, 2018 where Elon talked about how expensive it would be to compete. He writes, working at the cutting edge of AI is unfortunately expensive. I strongly suspect that compute horsepower will be necessary and possibly even sufficient to reach AGI. If historical trends are any indication, progress in AI is primarily driven by Systems Compute data infrastructure. The core algorithms we use today have remained largely unchanged from the 90s. Not only that, but any algorithmic advances published in a paper somewhere can be almost immediately reimplemented and incorporated. Conversely, algorithmic advances alone are inert without the scale to also make them scary. It seems to me that OpenAI today is burning cash and that the funding model cannot reach the scale to seriously compete with Google, an $800 billion company. If you can't seriously compete but continue to do research in open, you might in fact be making things worse and helping them out for free because any advances are fairly easy for them to copy and immediately incorporate at scale. A for profit pivot might create a more sustainable revenue stream over time and would with the current team, likely bring in a lot of investment. However, building out a product from scratch would steal focus from AI research. It would take a long time and it's unclear if a company could quote, catch up to Google scale and the investors might exert too much pressure in the wrong directions. The most promising option I can think of, as I mentioned earlier, would be for OpenAI to attach to Tesla as its cash cow. I believe attachments to other large suspects Apple, Amazon would fail due to an incompatible company DNA. Using a rocket analogy, Tesla already built the first stage of the rocket with the whole supply chain of Model 3 and its onboard computer and a persistent Internet connection. The second stage would be a full self driving solution based on large scale neural network training which OpenAI expertise could significantly help accelerate with a functioning full self driving solution in two to three years we could sell a lot of cars and trucks. If we do this really well, the transportation industry is large enough that we could increase Tesla's market cap and use that revenue to fund the AI work at the appropriate scale. I cannot see anything else that has the potential to reach sustainable Google scale capital within a decade and then finally an email from Wednesday, December 26, 2018 from Elon to Ilia, Greg Brockman and Sam Altman Subject I feel I should reiterate Body My Probability assessment of OpenAI being relevant to DeepMind Google without a dramatic change in execution and resources is 0%, not 1%. I wish it were otherwise. Even raising several hundred million won't be enough. This needs billions per year immediately or forget it. Unfortunately, humanity's future is in the hands of redacted. Then another redacted sentence and they are doing a lot more than this. Then another long redacted line and Elon ending. I really hope I'm wrong now. A bunch of sleuths took to AI to try to see if they could figure out what the redacted parts of that last email were. Pietro Scarano of Everardai writes, I used Claude3 to unredact this part from the OpenAI emails. What's wild is that they used a per word redaction, meaning each redaction length is proportional to the length of the words. So assuming context and word length, this is Claude's guess using the page source. So this suggestion for what the redaction reads is unfortunately, humanity's future is in the hands of Google. I believe DeepMind alone is spending north of $100 million per year and they are doing a lot more than this. They are building huge clusters and have a clear path to universal AI that can adapt to any problem and crush humans at every intellectual task while running on cheap hardware. Rohan Pandey did something similarly writing it's 5:46am Good morning. I spent the last few hours writing a character count constrained decoding algorithm for llama213b to de redact Elon's email to Ilya from 2018. Here's one of the completions it proposed that perfectly matches the length constraints of the redaction Rohan's findings Unfortunately, humanity's future is in the hands of Demis. Then a URL to DeepMind's blog post about AlphaFold and in the final redacted section nothing matters but the scale. And scale requires money. So regardless of the details, we have two redaction interpretations that put the line unfortunately, humanity's future is in the hands of as focused on someone associated with Google, either Google itself or Demis Hassabis at DeepMind. Now once again, there are wildly different interpretations of this. People from outside the AI space seem to be taking it as confirmation that Elon is kind of full of it and just wanted to be the CEO. Daro Bassano writes OpenAI responds to Elon Musk's lawsuit suing them for converting from a nonprofit to a for profit entity by sharing emails showing he knew of those plans since it was the only path to generate billions of revenue. The disagreement was he wanted to be CEO and merge it with Tesla. Jim Cramer screamed this morning that Elon only sued Open AI because he wants a quote show trial. Investor Vinod Khosla, who's been up in arms about this whole thing from the beginning, writes Elon Musk is being self righteous about the ills of prioritizing profit over benefit to humanity to the less naive. Timelines Show Elon left OpenAI and reneged on his word to, quote, cover whatever of the initial 1 billion anyone else doesn't provide. He reneged simply because he wanted to wrangle control for himself and for Tesla. However, on the other side, there are plenty of people who are honing in on the letter From Ilya on January 2, 2016 where they talk about becoming less open over time as the most important information from this I'll read that section again just so you remember it. Ilya writes, as we get closer to building AI, it will make sense to start being less open. The open in open AI means that everyone should benefit from the fruits of AI after it's built. But it's totally okay to not share the science, even though sharing everything is definitely the right strategy in the short and possibly medium term for recruitment purposes. Bindu Ready writes, this has been a power struggle of epic proportions. At stake was AGI and everyone wanted to control it. Open source was just a way to recruit people. Sadly, humanity's biggest failing has been its insatiable desire for power. History and mythology have multiple stories that outline these power struggles. If it is not evident by now, open source is the only way to benefit humanity and not a select few. Vancyonx writes, OpenAI was never about being open source. They all lied about being open to get more people in. Now this lawsuit feels like a personal grudge between Elon and Sam. Also, it makes everyone a villain here. Nathan Land said something similar writing Is it just me or does this blog post make all of the parties involved look bad? It's kind of sad because I'm a big fan of Elon, Sam and the OpenAI team. The word play around what open means and making it clear it is at least partially a marketing term for recruiting leaves a bad taste. And also it seems Elon wasn't entirely forthright about how things went down. On the flip side, while some are saying this makes everyone look bad. Others are saying everyone's position is kind of reasonable. Eacc founder Beth Jesos says okay, so Elon wanted to do an OpenAI partnership deal with Tesla in order to make OpenAI competitive with Google. Which makes sense because that type of deal is what Sam ended up doing with Microsoft instead. Seems reasonable from all sides. Question mark as I've said before, anything involving Elon right now ultimately becomes a Rorschach for how someone feels about Elon, which makes it a little bit difficult to get beyond that when you're trying to analyze an Elon related situation. I certainly think that these emails paint Elon's intentions and his perhaps motivations for this lawsuit in a very different light than some might have thought. It certainly adds evidence in the favor of a more personal, financial, and less altruistic motivation. But at the same time, it's fairly wild to me that OpenAI didn't anticipate how negative the response would be of them sharing that from 2016 they knew that they weren't really going to be open in the way that people assumed open meant. Perhaps that's a mistake of not realizing that people would hear in OpenAI and assume open source, not a more broad, generic, definable version of open. But in many ways it really does feel like this is the central issue that there is open in their name. As silly as that sounds, it feels like a really big deal. Putting a fine point on that is that Elon responded to OpenAI's post about this blog on X writing, change your name to Closed AI and I will drop the lawsuit. Crazy times continue in AI land, friends. But for now that will do it for the AI breakdown. Until next time. Peace. Sam.