Tech Brew Ride Home - Mon. 12/11 – Here Come The New Open-Source AI Startups

Episode Date: December 11, 2023

I told you the open-source AI startups were coming, and a big one is making waves today. Apple’s iPad lineup is hella confusing, but Mark Gurman says Apple is working to streamline that in the coming year. And let me give you the background on that whole Effective Accelerationist movement you might have been hearing about online. Sponsors: Miro.com/podcast CrashPlan.com/ridehome Links: Mistral, French A.I. Start-Up, Is Valued at $2 Billion in Funding Round (NYTimes) Nvidia Sees Vietnam as Potential Second Home, Reports Say (Bloomberg) Apple Is Working on Cleaning Up Its Confusing iPad Lineup (Bloomberg) China’s cyber army is invading critical U.S. services (Washington Post) This A.I. Subculture’s Motto: Go, Go, Go (NYTimes) 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 Monday, December 11th, 2023. I'm Brian McCullough today. I told you the open source AI startups were coming and a big one is making waves today. Apple's iPad lineup is hella confusing, but Mark German says they're working on that. And let me give you the whole background on that effective accelerationist movement you might have been hearing about online. Here's what you missed today in the world of tech. Let's open this week with an interesting raise, Paris-based mistral, AI has raised 385 million euro from A16Z, lightspeed, and others, sources say at around a $2 billion valuation. This is interesting for a couple of reasons. First, they're only eight months old.

They're only 22 people in total. But the founders are X-Meta and Google AI folk. Also, they're entirely an open source play. So when I've been telling you that the new hotness in AI at the moment is open source, well, this might be. be example 1A, quoting the times. The startup's value has increased more than sevenfold in just six months. In the summer, it raised a seed funding round of 105 million euros, about $113 million, that valued the company at about $260 million. Mistral builds technology that other businesses can use to deploy chatbot,

search engines, online tutors, and other AI-driven products. It is among a small group of companies, including the tech industries, giants, and a handful of startups that are building AI that could rival technology under development at OpenAI, the San Francisco startup that kicked off the AI boom last fall with the release of the chat GPT chatbot. Mistral is also among the companies that believe in sharing this technology as open source software, computer code that can be freely copied, modified, and reused, providing outsiders with everything they need to quickly build chatbots of their own. Rival companies like OpenAI and Google argue that the open source approach is dangerous and that the

raw technology could be used to spread disinformation and other harmful material. Mistral's fate has taken on considerable importance in France, where the finance minister has pointed to the company as providing the nation a chance to challenge U.S. tech giants. Europe has not produced many meaningful tech companies dating back to the dot-com boom and sees artificial intelligence as a field where it can gain ground. Investors are pumping money into other startups that believe in the open-source approach. Perplexity, founded last year by another group of top researchers has raised a new $70 million funding round that values that company at $500 million, a person familiar with the deal said. Investors there include IVP and Bessemer Venture

Partners. Mistral was founded by Timothy LeCroy and Guillaume Lample, who previously worked as researchers in the Paris AI Lab of Meta, the owner of Facebook and Instagram, and Arthur Mench, who was a researcher at Deep Mind, an AI lab that Google acquired in 2014 for $650,000. On Sunday, Mistral also released its technology as open source software, saying it performs at a level on par with Meta's technology, end quote. Yes, I heard a lot of folks chatting about this La Platform model last night. Here's a couple of choice tweets about this release. Lots of people claiming it outperforms GPT 3.5. Here's George Hatz. Google put out a press release and a fake demo. Mistral put out a torrent, end quote. And Alex McCaw. Looks like we might see an open source GPT4 level model much sooner than I thought, end quote. Remember last week when there were a bunch of headlines about Apple moving large chunks of its supply chain to India, but also Vietnam? Well, continuing that recent trend, Vietnamese media is reporting Nvidia's CEO Jensen Huang sees Vietnam as a potential second home for the Silicon Valley company and plans to open a design center there, quoting Bloomberg. Wang, speaking at a Hanoi Semiconductor and Artificial Intelligence Conference, hosted by a Vietnam Ministry Monday, said his company will open a design center, the new website VN Express, reported without giving details about the facilities technology.

NVIDIA has already invested about $250 million in Vietnam, the government's news website reported on Sunday, citing information from a meeting between Wang and Prime Minister Femmin Chin. Chin. Intel operates a chip assembly and test manufacturing facility in Ho Chi Minh City. Arizona-based Amcor Technology is building a $1.6 billion factory in northern Bakun province, while Synopsis and Marvel are establishing semiconductor design centers in Vietnam as well. Some 50 chip design companies have set up operations in Vietnam in recent years, end quote. Quick, off the top of your head, what are the differences between buying an iPad Pro, say, versus an iPad Air or versus a run-of-the-mill iPad? I mean, the iPad Mini is easy because it's obviously smaller.

But yeah, Apple's iPad lineup is currently confusing as I'll get out. And Mark German says, they're going to try to fix that in the new year. Quote, these days Apple sells five main iPads, the Pro, Air, and Mini, as well as the ninth and 10th generations of the regular iPad. In some cases, the products have just marginally different screen sizes and similarly. features. It's hard to know which one to pick. For instance, the 11-inch iPad Pro of today is only millimeters larger than the iPad Air, and the other differences are negligible for most people. The 10th-generation regular iPad, meanwhile, is nearly as good as the Air, but they're priced $150 apart. Apple is working to bring clarity to the iPad lineup. For starters, it wants to

reduce the confusion between the iPad Pro and the Air. The Pro is set for major changes, including an OLED screen, updated design, M3 chip, and revamped Magic keyboard attachment. That will make it unmistakably the highest-end model. In terms of screen sizes, the two models will be similar, but the pro will get you slightly more real estate. The iPad Air will come in a 10.9-inch and 12.9-inch configuration, while the Pro will be 11 and 13-inch. This mirrors the approach with the MacBook Air and MacBook Pro, where the latter model has a slightly larger screen. So the iPad Air will clearly be lower end than the Pro, but it also will be a notable improvement over the standard iPad. It will have two screen sizes and an M2 processor, making it superior to the

10th generation model, a product that isn't due to get upgraded until much later. As I reported this past week, the new Pro and Air models are coming around March. The new Magic Keyboard provides another differentiator for the iPad Pro. Apple isn't planning a new version of that accessory for the iPad Air. The new 12.9-inch model will stick to using the current Magic Keyboard. keyboard for that screen size. So if you want to get the best keyboard, you have another reason to spring for the pro. Now, the ultimate way to simplify the iPad would be dwindling the line down to just the pro in the air. But Apple needs a cheaper model to sell to the education market, or at risk losing more ground to Chromebooks. It also would be hard to ax the mini, despite

the proliferation of gigantic iPhones. Some people prefer a smaller tablet. One big question is if this simplification will be enough to reinvigorate the tablet category. In the short, term, new models at higher prices will help generate more revenue. But in the long run, an iPad comeback is anything but assured, especially when the device might have to compete with touchscreen MacBooks in a few years, end quote. More signs that China's military is ramping up its ability to disrupt key U.S. infrastructure via digital hacking, including targeting a Hawaii utility and a West Coast port in 2023, quoting the Washington Post. Hackers affiliated with China's People's Liberation Army have borrowed into the computer systems of about two dozen critical

entities over the past year, these experts said. The intrusions are part of a broader effort to develop ways to sow panic and chaos or snarl logistics in the event of a U.S.-China conflict in the Pacific, they said. Among the victims are a water utility in Hawaii, a major West Coast port, and at least one oil and gas pipeline, people familiar with the incidents told the Washington Post. The hackers also attempted to break into the operator. of Texas's power grid, which operates independently from electrical systems in the rest of the country. It is very clear that Chinese attempts to compromise critical infrastructure are in part to pre-position themselves to be able to disrupt or destroy that critical infrastructure in the

event of a conflict to either prevent the United States from being able to project power into Asia or to cause societal chaos inside the United States to affect our decision-making around a crisis, said Brandon Wales, Executive Director of the Department of Homeland Security, cybersecurity and infrastructure security agency or CISA. That is a significant change from Chinese cyber activity from seven to ten years ago that was focused primarily on political and economic espionage, he said, end quote. The disclosures to the post build on the annual threat assessment in February by the Office of the Director of National Intelligence, which warned that China, quote, almost certainly is capable of launching cyber attacks that would disrupt U.S. critical

infrastructure, including oil and gas pipelines and rail systems. If Beijing feared that a major conflict with the United States were imminent, it almost certainly would consider undertaking aggressive cyber operations against U.S. Homeland Critical Infrastructure and Military Assets worldwide, the assessment said, end quote. Finally today, since we've been talking about trends a lot today, and since we let off with that AI open source startup story, I wanted to hit up a thing that has been bubbling in the background that we've not been able to talk about much yet. If you're an extremely online person in the tech space, chances are you've run across the term EACC or effective accelerationism, which argues for open sourcing AI tools on one hand, but also says

that AI's benefits far outweigh their harms, so we should hurry up and get those benefits and opposes AI regulation. In other words, these are the full speed ahead with AI folks. And in the New York Times, Kevin Ruse gives us a summary of this. side of the evolving AI argument. Quote, effective accelerationism often shortened to E-S-ACC, pronounced E-AC, is a loosely organized movement devoted to the no-holds-barred pursuit of technological progress. The group believes that artificial intelligence and other emerging technologies should be allowed to move as fast as possible with no guardrails or gatekeepers standing in the way of innovation. The group formed on social media last year and bonded in Twitter spaces and group chats over

memes, late-night conversations, and shared scorn for the people they call D-cells and D-Comers, the people who worry about the safety of AI or the regulators who want to slow it down. It has moved offline, too, with parties and hackathons in the Bay Area and beyond. Effective accelerationism began as a cheeky response to an older, more established movement known as effective altruism that has become a major force in the AI world. EA, as the older group is known, got its start promoting a data-drable. approach to philanthropic giving, but in recent years has been worrying about AI safety and promoting the idea that powerful AI could destroy humanity if left unrestrained. The battle between the

EACs and the effective altruists is one of many quasi-religious schisms breaking out in San Francisco's AI scene these days, as insiders argue about how quickly the technology is progressing and what should be done about it. EAC prefers the all-gas, no-breaks approach. It's at favor open sourcing AI software rather than having it be controlled by big corporations. And unlike effective altruists, they don't see powerful AI as something to be feared or guarded against. They believe that AI's benefits far outweigh its harms and the right thing to do with such important technology is to get out of the way and let it rip. Some of the ideas EAC has adopted, like its opposition to regulation, are standard techno-libertarian gospel. Others resemble

tenants of older Silicon Valley subcultures like the transhumanists and the extropians, who also valued progress and resisted attempts to contain technology. The movement also borrows from the works of the British philosopher Nick Land, who wrote years ago that the accelerating forces of capitalism and AI would ultimately collide in a techno-capital singularity, a point at which technology would outstrip our ability to contain it. More recently, Mr. Land has fallen out of favor after endorsing far-right ideas about race and authoritarianism. In a manifesto posted online last year, EAC's founders, all of whom used inside-joke pseudonyms like Bayes Lord and Baysbeth Jaisos, described their goals in lofty, bombastic terms, writing that their

goal was to, quote, usher in the next evolution of consciousness, creating unthinkable next-generation life forms, end quote. Most people, of course, want to keep the light forms we already have, and critics of EAC, chafe at the idea that we should roll over and let the robots overtake us. Peter S. Park, an AI researcher at MIT, and the director of stakeout.a.I. and AI Safety Advocacy Group told me he considers EAC, quote, a dangerous, unaccountable ideology inspired by replacing humanity with AI, end quote. Initially, I wrote the movement off as a fringe novelty, a bunch of Twitter-addicted techies with persecution complexes, turning warmed over Ayn Rand into edgy memes. But a few months later, tech luminaries like Mark Andreessen, the co-founder of the venture capital firm Andresen Horowitz, started showing up in EIA. X's Twitter spaces and proclaiming that he too believed in effective accelerationism. Mr. Andresen's profile on X, the social network formerly known as Twitter, now includes EAC, and he listed

BASED BESOS-BES-LORD as two of his patron saints in the techno-optimism manifesto he published in October. Gary Tan, the president of the influential startup incubator Y Combinator, signaled his support for EAC. Sam Altman, the chief executive of OpenAI, replied to a based Beth Jaisos tweet and joked, quote, you cannot out-accelerate me, and the movement gradually broadened beyond AI with some leaders pushing for cryptocurrencies or nuclear fusion. Soon, the movement was gaining steam in Silicon Valley and officials in Washington were warning about its growing influence. It was a sure sign to the EAC crowd that they had trolled the right people. Last week, Forbes revealed that based Beth Jaisos was actually Mr. Verdon,

who now runs an AI hardware startup called X-Tropic. Mr. Verdon, who has had enough media exposure for one week declined to be interviewed for this column. His unmasking took some of the mystique out of EAC, but it didn't seem to dampen followers' enthusiasm. I interviewed several EAC supporters recently, ranging from early joiners to more recent converts. All of them praised the movement as a refreshing antidote to the pessimism of the AI safety crowd. I'm Jad Mossad, the chief executive of the AI coding startup, and an investor in Mr. Verdon's startup, told me that he liked EAC, quote, as a meme counterweight to all the AI doom and gloom, end quote. These are, of course, verdicts on EAC's vibes, not its ideas, some of which are still too extreme for many people to swallow.

Critics have pointed to the fact that some of EAC's leaders, including Mr. Verdon, seem to actually agree with the effective altruists that a rogue AI could wipe out humanity but aren't bothered by the idea since superhuman AI could represent a logical next step in evolution. And some have noticed that the movement has gotten more partisan and serious as it has grown. I liked it when it was an ironic counter movement instead of what seems to be transforming into an earnest libertarian movement, said Aidan Gomez, the chief executive of the AI company cohere. Even Grimes, who played an EAC party last month, has distanced herself from the movement saying in a post on X that she was, quote, DJing in enemy territory because I think healthy discourse is constructive, end quote.

I might have some big podcast related news for you tomorrow. It's another project I've been working on for a while that might launch any day now or might be delayed. But get ready. It's coming sometime this week. Got something cooking that I think you'll find useful. Talk to you maybe about this tomorrow.