Image by akinbostanci via iStockPhoto/Getty Images
AI and National Security with Mike Bloomberg
Your humble newsletter writer had a delightful meeting this week with America’s mayor and mensch Mike Bloomberg, where we discussed the opportunities and risks of AI and national security. Bloomberg is currently the chair of the Defense Innovation Board at the Pentagon, which if you haven’t followed, has been focused on the core building blocks of building data infrastructure across the department and lowering barriers to defense innovation. There’s still so much work to be done in readying America for the AI-driven warfighting of the future, but it’s heartening to know that engineers and entrepreneurs like Bloomberg have a seat at the table.
Yes, children deserve access to social media
The bedrock of human progress and ultimate flourishing is free-flowing information. The chaotic frenzy of discovery requires serendipitous juxtaposition of ideas, facts, and theories — both proven and falsified — to mix and meld into a new generation of thoughts in an infinitely-braided tapestry of intellectual threads. Fashions ebb and flow, ideas are neutered before dominating, and there’s a stochastic nihilism in the search for patterns on when discoveries will finally hit (as Jason Crawford of The Roots of Progressexplored, the components of a bike were available for millennia before it was finally invented).
Information is not enough of course — there have to be rigorous systems of debate coupled with a culture of skepticism to mold that clay of raw facts and fire them into the refined theories that deserve widespread respect (I’m reminded of Jonathan Rauch’s “The Constitution of Knowledge”). But the key ingredient — before anything else — is access to the wealth of human knowledge in all of its forms.
So it is with increasing alarm that governments are locking down information, and that there is a growing political consensus in the United States and throughout the world for blocking access to human expression.
The latest example came from Florida, which passed a law this week that would ban access to social media accounts by minors under the age of 14, even with parental consent. In the American education system, that age lies somewhere in the last two years of middle school. Under the law, 14- and 15-year-olds (freshman and sophomore high school students) would require parental consent to open an account.
This is a ban on accounts, not access. Children can still view public pages on social networks like Facebook and TikTok, they just can’t sign up and start to build their own networks or interact with many of the features of these platforms. Obviously, that restriction makes these platforms quite limited in their utility, since the ability to customize the content is pretty much the whole point.
In response to such research, politicians are increasingly moving away from regulating the product decisions of social media companies to outright bans. This week’s Florida law is an example of this new approach. The law identifies certain features of a social media service to be considered eligible for regulation under the law, including the ability to upload content, the length of time children spend on the site and whether the site features an algorithm with “addicting” features (auto-play videos, as an example). What’s different from past efforts is that these addicting features aren’t being targeted for regulation themselves, but account creation is.
Notably, kids under 14 are banned from having a social media account, even with parental consent. The idea is that the overwhelming demand by children for account access means that parents can no longer be considered responsible for managing their children’s time, and so the government must intervene on their behalf.
Like any technology, social media and the internet are products that afford a wide range of uses to people. The vast majority using TikTok or Facebook are using it for entertainment and to socialize with friends, often hours a day based on the latest time-use studies. There are indeed intellectual communities on these platforms, ranging from scientists and specialists discussing their fields to creative friends exchanging notes and videos about their latest work and soliciting feedback. Admittedly, this isn’t the mainstream use of these platforms, and so the argument is that there isn’t any real intellectual loss for young students that comes with an account ban. After all, you can visit Wikipedia, news services, and most research resources without running foul of this law.
Yet, removing the social dynamics of information and that “chaotic frenzy of discovery” is precisely what removing access to social media does. Consider services like X (Twitter) and Reddit, where communities can form organically across the web and connect people with similar interests. While both sites have their controversies, they are also home to flourishing communities of builders and scientists who share the knowledge of what they are discovering including to, yes, children who are curious about everything from rocketry to electrical engineering.
But our concept of social media still remains too constrained. Consider more professional repository hubs like GitHub or Lux’s own Hugging Face, which is the de facto platform for collaboration around machine learning models. These sites have “push notifications” that “inform a user about specific activities or events related to the user’s account.” Maybe they won’t trip the Florida law’s time requirement for children (“Ten percent or more of the daily active users who are younger than 16 years of age spend on average 2 hours per day or longer on the online forum, website, or application”) and so will be excluded, although I wouldn’t be willing to take that wager.
Right now, there are thousands of children dreaming about the potential of artificial intelligence, learning and thinking and building the future here and wanting to be part of one of the great transformational epochs that humanity has ever produced. Why would we want to do anything that might dampen that curiosity and passion through laws that force these children to log off and … do something else?
The challenge of regulating speech of course is that the U.S. government and subsidiary states and agencies can’t really pick what speech is allowed and what isn’t, and so they have to write laws focused on specific functions or time and place rules that will exceed court scrutiny. So in our rash attempt to stop children from cyberbullying each other on Instagram, we are also cutting out the many positive examples of social media that encourage children to engage with their own futures.
It’s discouraging to watch the growing censorship of social media in the United States and how that parallels similar moves in China. China has been on a campaign the last five years to drastically reduce the time children spend playing video games by mandating age-verified logins, banning children from playing games for more than an hour or two a week, and preventing spending on various business models like tips during livestreaming. The hope is that by controlling children’s time, the state can push children to focus more on schoolwork, pro-social behavior as well as ideological education initiatives.
In both countries, we are increasingly seeing children (and even parents!) as lacking agency, with the government empowered to direct and control their intellectual activities. Yet, that feels like precisely one of the root causes of the mental health problems that children are experiencing these days: the inability to control their world, decide on their course in life, and have any agency to do what they want to do. By taking choices away, we are only proving their sense of powerlessness and reinforcing the existential crises many children seem to be suffering from.
Children don’t need to be so mollycoddled, and certainly not by the state. Technology platforms like Minecraft and Roblox have shown that children can imagine spectacular worlds when afforded the right tools and constraints to express their creativity and share it with peers. After all, about half of the users on Roblox are estimated to be under the age of 12, out of more than 200 million monthly actives. Perhaps the answer is that most social media is actually reasonably decent, and just a small handful of admittedly popular platforms are the exception?
I don’t want to minimize the mental health crisis among teens, or the terrible situations that children can be led to on social media. But I do want to highlight that much like media technologies of the past, social media is today’s most critical set of tools for self-expression, learning, discovery, and maturation. Access to information and to forums for debate shouldn’t be roped off for the next generation. Human flourishing never begins with the censor’s red pen.
Lux Recommends
I haven’t read it (so I am treading in dangerous territory) but I did enjoy the review of Wayne Kalayjian’s new book “Saving Michelangelo's Dome: How Three Mathematicians and a Pope Sparked an Architectural Revolution.” From the review, “The book’s central claim is that, in the course of solving the vexing problem of cracks in the dome that threatened imminent collapse, the three mathematicians effectively founded the field of modern engineering. As obvious as it may seem to us now in an atmosphere of burgeoning interest in math and science, the idea of inviting mathematicians to look for creative solutions was by no means an obvious one in the 18th century.”
Our scientist-in-residence Sam Arbesman enjoyed this incredible story and interactive website from James Somers and Edwin Morris on “The Baffling Intelligence of a Single Cell” also known as E. coli chemotaxis. “We tend to think of a colony of something like E. coli as an undifferentiated evil goo, each bacterium identical to its neighbors. But people who’ve studied these organisms under the microscope observe a surprising amount of individual personality.”
A popular hit on Twitter and one that also hits close to home, Theo Baker pens an introspective feature from the West Coast in The Atlantic on “The War at Stanford” over the increasingly polarized life of the once-idyllic campus. How can an enlightened institution of debate still function when so many minds can never be changed?
Continuing The Atlantic’s takeover of Lux Recommends (and maybe Sam was reading my mind this week), but he highlights Ian Bogost’s latest column on “Social Media Is Not What Killed the Web.” “Because every click brought more delay, one clicked more deliberately. Browsers displayed visited links in a different color (purple by default, instead of blue). They still do this, but nobody cares anymore; using the OldWeb browser reminded me that those purple links helped you navigate a strange and arduous terrain. Yes, that’s where I meant to go, or Nope, already been there.”
Finally, because why not, in a time when media has been facing an evisceration of economics and status, The Atlantic has seemed to grab hold of a winning formula. “The Atlantic’s revenue grew 10% last year to close to $100 million, and the publication has nearly one million total subscribers, split equally among people with digital-only and print subscriptions.” As I wrote in “TechCrunch+ Termination,” subscriptions are the core of a healthy and sustainable future for media (as much as we all hate them).
That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.
Forcing China’s AI researchers to strive for chip efficiency will ultimately shave America’s lead
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
Right now, pathbreaking AI foundation models follow an inverse Moore’s law (sometimes quipped “Eroom’s Law”). Each new generation is becoming more and more expensive to train as researchers exponentially increase the number of parameters used and overall model complexity. Sam Altman of OpenAI said that the cost of training GPT-4 was over $100 million, and some AI computational specialists believe that the first $1 billion model is currently or will shortly be developed.
As semiconductor chips rise in complexity, costs come down because transistors are packed more densely on silicon, cutting the cost per transistor during fabrication as well as lowering operational costs for energy and heat dissipation. That miracle of performance is the inverse with AI today. To increase the complexity (and therefore hopefully quality) of an AI model, researchers have attempted to pack in more and more parameters, each one of which demands more computation both for training and for usage. A 1 million parameter model can be trained for a few bucks and run on a $15 Raspberry Pi Zero 2 W, but Google’s PaLM with 540 billion parameters requires full-scale data centers to operate and is estimated to have cost millions of dollars to train.
Admittedly, simply having more parameters isn’t a magic recipe for better AI end performance. One recalls Steve Jobs’s marketing of the so-called “Megahertz Myth” to attempt to persuade the public that headline megahertz numbers weren't the right way to judge the performance of a personal computer. Performance in most fields is a complicated problem to judge, and just adding more inputs doesn't necessarily translate into a better output.
And indeed, there is an efficiency curve underway in AI outside of the leading-edge foundation models from OpenAI and Google. Researchers over the past two years have discovered better training techniques (as well as recipes to bundle these techniques together), developed best practices for spending on reinforcement learning from human feedback (RLHF), and curated better training data to improve model quality even while shaving parameter counts. Far from surpassing $1 billion, training new models that are equally performant might well cost only tens or hundreds of thousands of dollars.
This AI performance envelope between dollars invested and quality of model trained is a huge area of debate for the trajectory of the field (and was the most important theme to emanate from our AI Summit). And it’s absolutely vital to understand, since where the efficiency story ends up will determine the sustained market structure of the AI industry.
If foundation models cost billions of dollars to train, all the value and leverage of AI will accrue and centralize to the big tech companies like Microsoft (through OpenAI), Google and others who have the means and teams to lavish. But if the performance envelope reaches a significantly better dollar-to-quality ratio in the future, that means the whole field opens up to startups and novel experiments, while the leverage of the big tech companies would be much reduced.
The U.S. right now is parallelizing both approaches toward AI. Big tech is hurling billions of dollars on the field, while startups are exploring and developing more efficient models given their relatively meagre resources and limited access to Nvidia’s flagship chip, the H100. Talent — on balance — is heading as it typically does to big tech. Why work on efficiency when a big tech behemoth has money to burn on theoretical ideas emanating from university AI labs?
Without access to the highest-performance chips, China is limited in the work it can do on the cutting-edge frontiers of AI development. Without more chips (and in the future, the next generations of GPUs), it won’t have the competitive compute power to push the AI field to its limits like American companies. That leaves China with the only other path available, which is to follow the parallel course for improving AI through efficiency.
For those looking to prevent the decline of American economic power, this is an alarming development. Model efficiency is what will ultimately allow foundation models to be preloaded onto our devices and open up the consumer market to cheap and rapid AI interactions. Whoever builds an advantage in model efficiency will open up a range of applications that remain impractical or too expensive for the most complex AI models.
Given U.S. export controls, China is now (by assumption, and yes, it’s a big assumption) putting its entire weight behind building the AI models it can, which are focused on efficiency. Which means that its resources are arrayed for building the platforms to capture end-user applications — the exact opposite goal of American policymakers. It’s a classic result: restricting access to technology forces engineers to be more creative in building their products, the exact intensified creativity that typically leads to the next great startup or scientific breakthrough.
If America was serious about slowing the growth of China’s still-nascent semiconductor market, it really should have taken a page from the Chinese industrial policy handbook and just dumped chips on the market, just as China has done for years from solar panel manufacturing to electronics. Cheaper chips, faster chips, chips so competitive that no domestic manufacturer — even under Beijing direction — could have effectively competed. Instead we are attempting to decouple from the second largest chips market in the world, turning a competitive field where America is the clear leader into a bountiful green field of opportunity for domestic national champions to usurp market share and profits.
There were of course other goals outside of economic growth for restricting China’s access to chips. America is deeply concerned about the country’s AI integration into its military, and it wants to slow the evolution of its autonomous weaponry and intelligence gathering. Export controls do that, but they are likely to come at an extremely exorbitant long-term cost: the loss of leadership in the most important technological development so far this decade. It’s not a trade off I would have built trade policy on.
The life and death of air conditioning
Across six years of working at TechCrunch, no article triggered an avalanche of readership or inbox vitriol quite like Air conditioning is one of the greatest inventions of the 20th Century. It’s also killing the 21st. It was an interview with Eric Dean Wilson, the author of After Cooling, about the complex feedback loops between global climate disruption and the increasing need for air conditioning to sustain life on Earth. The article was read by millions and millions of people, and hundreds of people wrote in with hot air about the importance of their cold air.
Demand for air conditioners is surging in markets where both incomes and temperatures are rising, populous places like India, China, Indonesia and the Philippines. By one estimate, the world will add 1 billion ACs before the end of the decade. The market is projected to before 2040. That’s good for measures of public health and economic productivity; it’s unquestionably bad for the climate, and a global agreement to phase out the most harmful coolants could keep the appliances out of reach of many of the people who need them most.
This is a classic feedback loop, where the increasing temperatures of the planet, particularly in South Asia, lead to increased demand for climate resilience tools like air conditioning and climate-adapted housing, leading to further climate change ad infinitum.
Josh Wolfe gave a talk at Stanford this week as part of the school’s long-running Entrepreneurial Thought Leaders series, talking all things Lux, defense tech and scientific innovation. The .
Lux Recommends
As Henry Kissinger turns 100, Grace Isford recommends “Henry Kissinger explains how to avoid world war three.” “In his view, the fate of humanity depends on whether America and China can get along. He believes the rapid progress of AI, in particular, leaves them only five-to-ten years to find a way.”
Our scientist-in-residence Sam Arbesman recommends Blindsight by Peter Watts, a first contact, hard science fiction novel that made quite a splash when it was published back in 2006.
Mohammed bin Rashid Al Maktoum, and just how far he has been willing to go to keep his daughter tranquilized and imprisoned. “When the yacht was located, off the Goa coast, Sheikh Mohammed spoke with the Indian Prime Minister, Narendra Modi, and agreed to extradite a Dubai-based arms dealer in exchange for his daughter’s capture. The Indian government deployed boats, helicopters, and a team of armed commandos to storm Nostromo and carry Latifa away.”
Sam recommends Ada Palmer’s article for Microsoft’s AI Anthology, “We are an information revolution species.” “If we pour a precious new elixir into a leaky cup and it leaks, we need to fix the cup, not fear the elixir.”
I love complex international security stories, and few areas are as complex or wild as the international trade in exotic animals. Tad Friend, who generally covers Silicon Valley for The New Yorker, has a great story about an NGO focused on infiltrating and exposing the networks that allow the trade to continue in “Earth League International Hunts the Hunters.” "At times, rhino horn has been worth more than gold—so South African rhinos are often killed with Czech-made rifles sold by Portuguese arms dealers to poachers from Mozambique, who send the horns by courier to Qatar or Vietnam, or have them bundled with elephant ivory in Maputo or Mombasa or Lagos or Luanda and delivered to China via Malaysia or Hong Kong.”