How should one design survey experiments for maximal impact in economics? Here's my slide deck from a recent workshop in Uppsala. Thanks to everyone attending for making it a great experience @EconomicsUU drive.google.com/file/d/1yN4…

It would be hard to dream up a system better designed to inhibit child bearing than today’s academia. Endless grad school, pre and postdoc, no location stability, quantity not quality evaluation, etc. yet they wonder why so few women do it

I disagree with this decision and I don't like it. But also... HOW DID ANTHROPIC NOT SEE THIS COMING‽ It is *the* obvious response to "this is too dangerous for anyone except us to use", since that relies on a premise ("we are uniquely good") that almost no-one agrees with.

There’s a big, under-appreciated reason why people may have very different experiences and opinions about using AI for work — are they using it for tasks they’re already an expert at, or tasks they can’t do themselves? The former leads to a *growth cycle* and the latter leads to a *dependence spiral*. When I use AI to do something I’m an expert at, like coding, I treat it as a tool. I can build quickly, maintaining an understanding of the code, knowing that if necessary, I can fix the code myself. It feels empowering. It frees up my time to think about the complex, judgment-oriented parts of software engineering that I can’t or won’t delegate to AI. That means my own skills improve rapidly, and I get to climb the ladder of complexity and develop higher-level skills, much more so than when I write the code myself. I feel in control. I can lock in and achieve a flow state — when AI is working, I’m reviewing, building understanding, and planning the next steps. I never get the feeling that the tool is about to replace me. This is the growth cycle. (Of course, the growth cycle is not automatic. I still need to exercise agency to use AI responsibly. But it’s the same challenge with any productivity-enhancing technology, and those who’ve navigated such transitions before are well-equipped to navigate it with AI as well.) On the other hand, if I use it for tasks I don’t understand and haven’t learned to perform myself, I have no choice but to treat it as a superintelligence. If something breaks, the best I can do is ask AI to fix it and hope for the best. I generally can’t evaluate the quality of the output myself. The only way to find out if it's any good is if and when the work is ultimately reviewed by an actual expert. The experience is confusing, unsettling and disempowering. And forget about flow state. By over-relying on AI, I risk losing whatever skill I had at the task in the first place, even if it boosts productivity in the short term. This is the dependence spiral. It’s no wonder that entry-level workers and students preparing to enter the workforce find themselves in a bind. To compete with the AI-enabled productivity of more seasoned workers, they must adopt AI themselves, but doing so risks the dependence spiral. I have some thoughts on solutions that I will share in later posts, but I think having a clear diagnosis of the problem is a useful first step.

I applaud Piketty et al for embracing artificial intelligence.

Are you thinking about running an experiment to study racial discrimination? Are you going to use names to signal race? If so, I have a new tool for you! [link below] This tool helps you to choose the names you want to use in your study based on... 1.) the strength of the racial signal you want to send 2.) whether you want to hold other factors (e.g., class) constant In the past, names have been chosen in an ad hoc basis. We're trying to add the heft of name pre-testing to the study of racial discrimination.

I don't know what Piketty, Stiglitz, and co. are smoking. Global poverty rates have never been lower. Progress on basic global health and wellbeing measures has been amazing over the past few decades. "End of the road"?!? Come again!?! theguardian.com/commentisfre…

This paper tries to do much. Please split it into three papers so we can reject each one individually.

They didn’t mean pause AI research, they meant pause *your* AI research

Brilliant idea! Next up: Apple randomly reboots your Mac if you're building competing tech, Gmail silently edits your email if you mention rival platforms, and Tesla Autopilot swerves if it detects you're working on self-driving cars. All in the name of safety, of course. Because malicious actors controlling the world’s operating systems, inboxes and cars would be extremely dangerous!

Also, those who focus on using AI to help improve the skills of themselves and their teams will be diamonds in great demand, since they will be the rare A players in a sea of mediocrity.

Our @Nature comment this week on the use of AI in maths and theoretical physics - and why the community should embrace it! Authors @London_Inst & @GoogleDeepMind. First draft 8 months ago but edited many times as the field steamed ahead! Free-to-read link at the end of 🧵1/

Massive output uptick due to agentic AI. Complete flat adoption.

LLMs are intelligent, but they clearly aren't conscious? You could (incredibly slowly) run an LLM by hand by doing the matrix calculations with a pen and paper, would that be conscious? There is zero difference between doing that and doing it on a GPU except its faster.

I too am enjoying Hunter Biden being a great poster, but some people are a little too surprised that the son of a President is smart, well-informed, and good at communicating, as if they think addiction only happens to idiots

A remarkable video: Putin, an obscure 1991 Leningrad bureaucrat, warning against the return of totalitarian rule to Russia—paired with his reaction to those words as President in 2002. The irony writes itself.

Demis Hassabis recommendation for college students. He’s still do STEM, math and computer science. Expertise in those fields will help better leverage AI for at least next decade. Those in non-techical majors, really “lean in to” using latest models. AI labs spending so much time creating new models that they’ve only “scratched the surface” of what the models can actually do (huge “capability overhang”). And expertise in any field can be turbocharged by smart use of AI. “Double down on your own agency. The future is still to be written. Don’t listen to anyone that says it’s not.”

Really enjoyed this episode. Thanks to @dwarkesh_sp and @pawtrammell for the conversation. What I hope that I was able to convey that it is incredibly difficult to make predictions when there is so much uncertainty: there is not just uncertainty around the parameters, but even what model to use in the first place. In my view, the best application of economics to our current moment is not trying to individually forecast scenarios 5 or 10 years out (though aggregate forecasts are useful). There is way too much uncertainty at every level of the exercise. It’s to model important scenarios and work our way backwards: start with a potential scenario that are important to consider and then derive the conditions under which it can arise. This not only allows you to potentially rule out a very intuitive-sounding scenarios because the conditions required are implausible. It also points to data that you need to track which you were not considering before. Eg latent demand for human involvement, substitution between AI and human interaction, task bundling inside jobs, AI bottlenecks, and whether AI looks more like electricity or social media. This is the type of data I’m working to collect, and I know other teams are too. The last point is particularly important. To quote Demis Hassabis, we are potentially at the foothills of the singular. As economists we have the responsibility to guide that transition with both humility and the best information we can gather.

Building apps has never been easier. With Sites, Codex can turn your work, ideas, and plans into an interactive website or app your team can explore, use, and share with a URL. Rolling out to Business and Enterprise plans, before expanding more broadly.

Super excited to share joint work with @axiommathai that kicks off a broader project of formalization in economics. Aumann's celebrated theorem says we can't "agree to disagree." But what does that actually mean – formally? 👀

If you're not using ChatGPT Pro to speculate about history and war, you're missing out. What are the most ambitious seaborne invasions in history? Was the Grande Armée the best war force in history? How likely was it that Napoleon could have succeed with his plans to invade Britain? What would Hitler's chances be if he focused on breaking the UK before going to war with Russia? Can Taiwan sustain an attack from China? Can we play a war game where you're China and I'm Taiwan?