Tech Brew Ride Home - Fri. 03/01 – Elon Sues Sam Altman And OpenAI

Episode Date: March 1, 2024

Elon is suing Sam Altman and OpenAI. Meta continues to get out of the news business, especially when it’s being pushed. A new AI deepfake supertool. And, of course, the Weekend Longreads Suggestions. Links: Elon Musk sues OpenAI over AI threat (Courthouse News Service) Meta to Wind Down Its News Feature in the US and Australia (Bloomberg) Microsoft introduces Copilot AI chatbot for finance workers in Excel and Outlook (CNBC) Alibaba’s new AI system ‘EMO’ creates realistic talking and singing videos from photos (VentureBeat) Weekend Longreads Suggestions: Wait. Why is Reddit losing so much money? (Business Insider) The 9-Month-Old AI Startup Challenging Silicon Valley’s Giants (WSJ) What a Major Solar Storm Could Do to Our Planet (The New Yorker) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech meme right home for Friday, March 1st, 2024. I'm Brian McCullough today. Elon is suing Sam Altman and Open AI. Meta continues to get out of the news business, especially when it's being pushed, a new AI deep fake super tool and, of course, the weekend long read suggestions.

Here's what you missed today in the world of tech. Well, Silicon Valley has gotten the eating popcorn jiff lawsuit of its dreams that it deserves. Pick your metaphor, I guess. Elon Musk is suing open AI and Sam Altman, accusing them of betraying an agreement from OpenAI's founding to develop AI toward the, quote, benefit of humanity over profits. The suit was filed Thursday night in San Francisco Superior Court with Musk, claiming that OpenAI's tight relationship with Microsoft runs in direct opposition to the company's founding documents, quoting the suit. OpenAI, Inc. has been transformed into a closed-source, de facto subsidiary of the largest technology company in the world, Microsoft. Under its new board, it is not just developing but is actually refining an AGI to maximize profits for Microsoft rather than for the benefit of humanity, end quote.

A couple of interesting things there. First, he did not name Microsoft in the suit. Instead, OpenAI and CEO Sam Altman and President Greg Brockman are named as co-defenders. Though the implication here is that Microsoft is profiting off of what, if you're reading the language of the suit carefully, Elon claims, is already. AGI or artificial general intelligence. See, the main way OpenAI could be in breach of its founding documents is if it's profiting from AGI. If AGI has been achieved, no one is allowed to profit from it. But to get there, to get to the AGI, the idea was the profits were to be used to develop the tech that would become AGI. So Musk is basically saying, OpenAI already has achieved AGI. They're lying to you and profiting on it. so I'm suing to stop this. Quoting courthouse news service. Musk brings claims including breach of contract, breach of fiduciary duty, and unfair business practices against OpenAI and asked for the

company to revert back to open source. Artificial General Intelligence, a type of AI developed to autonomously perform on the cognitive level of humans, has been Open AI's main goal and is demonstrated, Musk says, in its GPT4. GPT4 was released in March 20203, but according to Musk, remains a closed model in contrast to previous iterations, a move driven by commercial considerations rather than the interest of humanity. The internal details of GPT4 are known only to OpenAI and on information and belief to Microsoft. GPT4 is hence the opposite of OpenAI, Musk says in the suit, and it is closed for proprietary commercial reasons. Microsoft stands to make a fortune selling GPT4 to the public, which would not be possible if OpenAI, as it is required to do, makes the technology freely available to the public. Contrary to the founding agreement,

defendants have chosen to use GPT4 not for the benefit of humanity, but as proprietary technology to maximize profits for literally the largest company in the world, he adds, end quote. The suit also mentions the late last year firing and then reinstatement of Altman as CEO and then the firing of the board that fired Altman. Musk basically asserts that the result of that drama was essentially a board no longer of scientists and researchers, but of P. people pliant to Altman who want to pursue business ends. Quote, open AI inks once carefully crafted nonprofit structure was replaced by a purely profit-driven CEO and a board with inferior technical expertise in AGI and AI public policy.

The board now has an observer seat reserved solely for Microsoft, end quote. So in summation, Musk is claiming Microsoft's de facto leverage on Altman and the board incentivizes them to keep OpenAI from declaring AGI has been achieved. in order to continue to profit from it. It would be sadly perfect if Microsoft is actually sitting on AGI and what they've done with it so far is to check notes, add chatbots to Excel. Meta is planning to deprecate the Facebook news tab in the U.S. and Australia in early April and stop signing deals for traditional news content in those countries, quoting Bloomberg.

People will still be able to view news on Facebook in. in feed in these countries and publishers will continue to have access to their Facebook accounts and pages where they can post their news, article links, and content. Meta said in a post, the change also doesn't affect existing agreements that Facebook has with news publishers at least until those deals expire. Lawmakers around the world have pushed Meta to compensate publishers for the stories that appear on the social network, but Meta says that fewer users are turning to Facebook for news anymore. The company previously stopped promoting news content for users in the UK, Germany, and France. The number of the number of the number of

of people using Facebook news in Australia and the U.S. has dropped by over 80% last year, according to the blog post. Meta has also been avoiding making recommendations for posts about politics or political issues. Meta previously suspended all links to news content on Facebook and Instagram and Canada to avoid making payments under the country's Online Streaming Act, end quote. This news has caused Australia's prime minister to threaten action against Meta. News Corp says that the move threatens the viability of small media groups. quoting the FT. Anthony Albanese, Australia's prime minister, says he was very concerned about the impact of meta's move. His government said it would, quote, work through all available options under

the news media bargaining code. Journalism is important, and the idea that research and work done by others can be taken for free is simply intolerable. Albanyes said in an interview with the Australian newspaper, we will respond in the national interest, he said, describing any situation where a company's profits from another's work as unfair and not the Australian way. One option for the Australian government would be to designate meta, which would force the company into arbitration over payments to media companies or face financial penalties. Meta and Google struck deals in 2021 to pay Australian media companies, including News Corp, newspaper publisher 9 Entertainment, and a host of smaller websites, broadcasters, and regional newspapers,

an aggregate total of more than 200 million Australian or 130 million dollars US a year, end quote. I mentioned this at the end of the first segment as a bit of a joke, but to be fair, this is more than a glib joke. Microsoft has launched co-pilot for finance into public preview, helping users reconcile data in Excel, speed up the collections process in Outlook, and more. In short, do accounting and financial stuff, moving from coding to calculating, quoting CNBC. Many business software providers, including HubSpot and Salesforce, have been working to supercharge existing products with generative artificial intelligence in the hope of making clients more efficient. The typical company comprises a variety of groups in which employees perform specialized work. We want every one of these departments to be enabled and enriched with

a co-pilot, Charles Lamana, a Microsoft corporate vice president said in an interview with CNBC in San Francisco on Wednesday. Microsoft already has a co-pilot for general purpose industrial use in office applications, and it has released co-pilots design for sales and customer service workers. The co-pilot for finance will initially run a variance analysis, reconcile data in Excel, and speed up the collections process and outlook. The software can draw on information stores, in SAP and in Microsoft Dynamics 365. Additional features will come to the finance co-pilot later this year, LaMana said. The Japanese advertising agency, Dentsu, will use the co-pilot for financial tasks, Lamana said. Microsoft said its own finance department provided input into the development of the new

co-pilot and that it's seen some early benefits from using it. Comparing data taken from different systems is, quote, something every finance team on the planet does a lot of, says Corey Hurnchik, modern finance lead in Microsoft's office of the chief financial officer, a couple of thousand on a financial planning and analysis team each spend one to two hours doing reconciliation each week. And with the new co-pilot, that takes more like 10 to 20 minutes per week, he said. The idea is to help these employees spend fewer hours on tedious tasks and provide time for more engaging work that can contribute more to the company. But Microsoft's finance employees aren't required to use the new co-pilot, Microsoft said,

end quote. On the AI video and possibly AI deep fake beat, Alibaba researchers have detailed EMO or Emote Portrait Alive, an AI system that creates a realistic talking or singing video from a portrait photo and an audio file, quoting Venture Beat. The system described in a research paper published on Archive is able to create fluid and expressive facial movements and head poses that closely match the nuances of a provided audio track. This represents a major advance in audio-driven talking head video generation, an area that has challenged AI researchers for years. The EMO system deploys an AI technique known as a

diffusion model, which has shown tremendous agility for generating realistic synthetic imagery. The researchers trained the model on a dataset of over 250 hours of talking head videos curated from speeches, films, TV shows, and singing performances. Unlike previous methods that rely on 3D face models or blend shapes to approximate facial movements, EMO directly converts the audio waveform into video frames. This allows it to capture subtle motions and identify specific quirks associated with natural speech. According to experiments described in the paper, EMO significantly outperforms existing state-of-the-art methods on metrics measuring video quality, identity preservation, and expressiveness. The researchers also conducted a user study that found the videos generated by EMO

to be more natural and emotive than those produced by other systems. Beyond conversational videos, EMO can also animate singing portraits with an appropriate mouth shape and evocative facial expression synchronized to the vocals. The system supports generating videos for an arbitrary duration based on the length of the input audio. The EMO research hints at a future where personalized video content can be synthesized from just a photo and an audio clip. However, ethical concerns remain about potential misuse of such technology to impersonate people without consent or spread misinformation. The researchers say they plan to explore methods to detect synthetic video, end quote. Time for the weekend long read suggestions. First up, something I've wondered about myself.

as it prepares for its IPO, and now that we can take a look at its financials, why, almost 20 years into life, why does Reddit not make money? Or as Business Insider titles this piece, why is Reddit losing so much money? Quote, last year, according to Reddit's newly filed IPO documents, the company generated $804 million in revenue and lost nearly $91 million. That's better than the year before, when it lost $159 million on revenue of $667 million. But it's still a lot of money to lose selling a free product, right? Last year, Reddit's spending on research and development, which it says is money spent primarily on, quote, engineers and other employees engaged in the research, design, and development of new and existing products, total of $439 million, and

eye-popping 55% of its revenue. By comparison, when Facebook went public in 2012, R&D was 10% of its revenue. Last year, when it was building things like virtual reality goggles, that number had bumped up to 29%. When Twitter went public in 2013, R&D was 44% of revenue. And at the end of 2021, the last year it filed a public statement before Elon Musk bought the company, that number was down to 25%. Bear in mind, this was a company that both Musk and Twitter management thought was overstaffed. Those numbers puzzle me. Reddit works great for the people who love it, and there are a lot of them. But one of the reasons it works great is that it's a pretty bare-bones product that does what it's supposed to. It lets people post something they're interested in and then lets other people comment on it.

That's it. That's the whole thing. What am I missing? I asked Reddit comms for comment, but they declined, citing the company's quiet period before the IPO. The best argument I can make in its defense is that Reddit is still adding a lot of users, and that more users equals more ad money, end quote. Then we've spoken a lot about them recently, so Grock this recent profile in the Wall Street Journal of mistral, that nine-month-old French AI company that is going toe-to-to-toe with the big dogs. Quote, this time last year, Arthur Mensch was 30, still employed at a Google unit here, and artificial intelligence had just started to take off in the public consciousness as something more than science fiction. Since then, so-called generative AI that can converse

and possibly reason like humans has become the most talked-about technology breakthrough in decades, and the startup Munch left Google to launch, now all of nine months old, is valued at slightly more than $2 billion. Mench's startup called Mistral AI is challenging the conventional wisdom that the winners of the AI race will emerge from among the tech industry's U.S. giants. Mench, who founded the company with two engineering school friends, doesn't think enormous scale is essential or that the U.S. will necessarily dominate. I've always regretted that there was no big tech in Europe, Mench, 31 said, at Mr. AI's Paris office, I think this is our chance to become one.

Mench, who started in academia, has spent much of his life figuring out how to make AI and machines. learning systems more efficient. Early last year, he joined forces with co-founders Timothy LeCroy, 32, and Guillaume Lample, 33, who were then at Meta Platform's Artificial Intelligence Lab in Paris. Together, they are betting that their small team can outmaneuver Silicon Valley Titans by finding more efficient ways to build and deploy AI systems, and they want to do it in part by giving away many of their AI systems as open source software. We want to be the most capital-efficient company in the world of AI, Munch said. That's the reason we exist. On Monday, Mistral unveiled a new AI model called Mistral Large that MENCH says can perform some reasoning task comparably with

GPP4, OpenAI's most advanced language model to date, and Gemini Ultra Google's new model. Munch said that his new model costs less than 20 million euro, the equivalent of roughly $22 million to train. By contrast, OpenAI chief executive Sam Altman said last year after the release of GPT4 that training his company's biggest models cost, quote, much more than 50 million to $100 million, end quote. And finally, from my dad, doom-scrolling file, one of the biggest black swan threats still out there, one that I've worried about for a long time, the New Yorker has a huge piece looking at the threat posed by solar storms. Quote, regular earth-based weather is such a fundamental part of our lives that we are almost always aware of it,

and very often obsessed with it, it is the subject of everything from idle chit-chat to impassioned political debate. By contrast, most people have no idea that there is. is weather in outer space, let alone what its fluctuations might mean for our planet. That's because, unlike everyday weather, you can't experience space weather directly. It doesn't make you hot or cold, doesn't flood your basement or take the roof off your home. In fact, until the 19th century, it had almost no appreciable effect whatsoever on human activity. Then came a series of scientific revolutions that made certain technologies from electricity to telecommunications central to our lives. Only later did we realize that those technologies are vulnerable to the effects of weather and outer space.

The potential consequences are as sweeping as our technological dependence. In 2019, the Federal Emergency Management Agency surveying the landscape of possible disasters concluded that only two natural hazards have the capacity to simultaneously affect the entire nation. One is a pandemic. The other is a severe solar storm. On March 13, 1989, a coronal mass ejection struck the, Earth. Within 90 seconds, transformers on the Quebec power grid malfunctioned, dozens of safety mechanisms failed, and the entire grid shut down, leaving almost a quarter of the population of Canada in the dark.

That geomagnetic storm, which also triggered outages in the UK and Sweden, destroyed a transformer at a nuclear power plant in New Jersey and caused at least 200 other issues on the North American grid alone. This storm was strong, but not exceptionally so, based on magnetometer readings, oral latitudes and other fingerprints left behind by solar storms, scientists now believe that at least three storms in the past 150-odd years, the Carrington event and others in 1872 and 1921, were roughly in order of magnitude more powerful. All three of those storms took place before the power grid existed. The question that troubles space weather experts and divides them is what will happen the next time a comparable one strikes. Some people think that the Quebec event was a wake-up

call the perfect-sized storm, really large, large enough to teach a lesson without being large enough to cause a catastrophe. But per the NAS report, any gains following the Quebec storm were offset by trends in America's bulk power system, which came to rely on ever-larger amounts of power traveling through ever-longer transmission lines. A study commissioned by the federal government and summarized in the report found that a storm the size of the 1921 event would cause large regions of the grid to fail, with impacts that, quote, would be of unprecedented scale and involve populations in excess of 130 million. Close to half of all Americans.

The report estimated the cost of a storm like that as $1 trillion to $2 trillion during the first year alone with recovery times of four to 10 years, end quote. So I'm not exactly sure if I have a bonus episode for you this weekend. We were supposed to record one yesterday that got postponed to next week. I'm scheduled to record one right after I hit publish on this today, but it might be embargoed until a later date. Still not sure yet.

So I don't know. If you get a bonus episode in your feed tomorrow, surprise? Otherwise, talk to you on Monday.