How to shape the technology that shapes us
Project Liberty
If there is a meta-takeaway, it is that humans have always found a way to reconnect and come back together. After a season of aloneness, history shows that our greatest breakthroughs are born from profound moments of unity.

People Want Control of Their Data
Project Liberty
According to research from Project Liberty Institute released last month, there’s broad global sentiment that online companies and platforms have too much personal information about everyday people, and those people don’t have the control they desire to manage their personal digital privacy.

Social Networks, Not Social Media
Clubs, connections, and community.
Glimpse into Glass
Glass may look similar to other services on the surface (there are conventions), but it’s built on fundamentally different principles. For one, we don’t have follower counts. In fact, we don’t show counts at all. When you appreciate a photo, it’s not a performative act broadcast to the larger network; it’s a private acknowledgment to the creator. We emphasize comments and discussion, modeling thoughtful engagement through design and community norms. Over time, our members have embraced this approach, despite years of conditioning by other platforms. It’s been glorious to witness. And, of course, the big one: Glass operates as a paid membership-based community.

Davos 2025 At Future House: Leaders Gather To Take On Social Media’s Impact On Gen Z

Discussing free speech w/ Hamish
A recording from Taylor Lorenz’s live video

Is DeepSeek the new DeepMind?
AI supremacy isn’t just about compute or U.S. leadership, it’s about how you work to make models more efficient and improve their accessibility for everyone.

What’s behind the DeepSeek freakout?

In tech terms, what freaked everyone out about DeepSeek’s R1 model is that it replicated — and in some cases, surpassed — the performance of OpenAI’s cutting-edge o1 product across a host of performance benchmarks, at a tiny fraction of the cost.

China pulls a Silicon Valley on… Silicon Valley

Given America’s strict export controls on high-end chips, analysts are still trying to figure out the exact details of how DeepSeek managed to match the top American models on a shoestring. DeepSeek itself claims it was done with ingenuity rather than fancy equipment — based on good old-fashioned optimization and low-level programming grunt work.

On Private Governance
Forging a new path

I am pleased to inform you that I will be taking on a consulting role with Fathom, a new nonprofit working on AI governance and organizer of the recent (and excellent) Ashby Workshops, as a fellow. In that capacity, I will be researching private governance of AI—that is, standards, best practices, and other mechanisms for the governance of frontier AI that are developed outside of formal governments (though with oversight from government)

Novus Ordo Seclorum
Reflections on DeepSeek
It’s not that we understand something that others are not smart enough to get. It’s all pretty straightforward, actually. Mostly I think we are just early. Early to the insight that AI is going to overturn countless things about the status quo, including at least some things that most people like, or at least find comfortably familiar. Early to the idea that this technological revolution will be intense, chaotic, and filled with uncertainty. Early to the excitement, and early, I must admit, to the anxiety. Early to the knowledge that we are stepping into a novus ordo seclorum—a new order of the ages. And early to feeling all of this in our bones, rather than just understanding it in some abstract intellectual way

Teaching Machines to Reason
Explanation of DeepSeek’s R1 paper
Perhaps the most exciting part of their research came next. The team discovered that they could create smaller, more efficient versions of their system through a process called distillation. Think of it like creating a concentrated essence of the original model’s capabilities.

China’s DeepSeek Adds a Weird New Data Point to The AI Race
V3 and R1 are Impressive Work, With Many Implications – but not “China Has Caught Up”

15 Observations About Artificial Intelligence
Thoughts about DeepSeek, ChatGPT, and the future of humanity.

What fully automated firms will look like
Everyone is sleeping on the *collective* advantages AIs will have, which have nothing to do with raw IQ: they can be copied, distilled, merged, scaled, and evolved in ways humans simply can’t.

The Lesson of DeepSeek
Tech Changes From the Bottom Up

The Coming Biotech Era
A New Era of Abundance
Now add AI. If you’re looking for a Ph.D research project, you can query AlphaFold and come up with a proposal quickly. AI will get you halfway through the grant and paper writing process. All you need to do then is your experiment, and new techniques like Cryogenic Electron Microscopy (Cryo-EM) speed that process along.

The First AI Bust is Here
Sell Hardware, Wait to Buy Software

The race for “AI Supremacy” is over — at least for now.
Decades of government kowtowing to Big Tech has thus far failed to produce a decisive victory
Gary Marcus
The race for “AI Supremacy” is over, at least for now, and the U.S. didn’t win. Over the last few weeks, two companies in China released three impressive papers that annihilated any pretense that the US was decisively ahead.

I spent two days testing DeepSeek R1
DeepSeek’s R1 model is almost as good as OpenAI’s o1—and much cheaper.

I don’t believe DeepSeek crashed Nvidia’s stock
But let’s talk about DeepSeek anyway.

AI Agents: RIP Autonomy
Emerging AI Governance Challenges
Luiza Jarovsky
Today, I want to discuss the agentic wave and explore why I call it the beginning of the death of human autonomy as we know it, especially through the lens of AI governance.

Some simple lessons from China’s big AI breakthrough`
Preventing LLM technology from spreading is a futile effort, but export controls can still work.

Why building big AIs costs billions – and how Chinese startup DeepSeek dramatically changed the calculus

Ambuj Tewari, University of Michigan

A machine learning expert breaks down where the money goes in building big AIs, and how DeepSeek found ways to do it far more cheaply.

AI agents’ promise to arrange your finances, do your taxes, book your holidays – and put us all at risk

Uri Gal, University of Sydney

AI systems that can autonomously make decisions on our behalf will be a huge time saver – but we must deploy them with care.

Silicon Valley’s bet on AI defence startups and what it means for the future of war – podcast

Gemma Ware, The Conversation
Political theorist Elke Schwarz talks to The Conversation Weekly podcast on her new research about venture capital investment into defence start-ups.

Could AI replace politicians? A philosopher maps out three possible futures

Ted Lechterman, IE University
From impartial debate mediators to a full-blown ‘algocracy’, we have to think carefully about how AI will impact politics.

Opening the black box: how ‘explainable AI’ can help us understand how algorithms work

David Martens, University of Antwerp; Sofie Goethals, University of Antwerp

AI systems can appear to be black boxes – often, even experts don’t know how systems reach their conclusions. The nascent field of “explainable AI” aims to address this problem.

Medieval theology has an old take on a new problem − AI responsibility

David Danks, University of California, San Diego; Mike Kirby, University of Utah
Autonomous AI is still designed by people − so who or what is really responsible for its actions? For centuries, theologians have posed similar questions about mankind and God.