New Anthropic research: How AI agents shape experts' judgment calls
When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.
3215534.3K780K
Christian Catalini✓
@ccatalini
Founder w/roots in academia. Founder @MIT Cryptoeconomics Lab. Past: Co-Founder & Chief Strategy Officer, Lightspark. Co-Creator, Libra. Head Economist, Meta.
6K Following26.6K Followers11.6K Posts
Christian CataliniRT @PChuzeville: Almost every dollar of AI safety funding today goes to understanding, testing, containing or defending the model itself, on the assumption that the danger is a model that misbehaves. But safety money only covers the first half of an AI output’s life, and almost no funder has dug into these questions: - Did anyone actually check this output? - Whose judgment is it, the person's or the agent's? - When it turns out to be wrong, who pays? Funders mistakenly leave them out for a few reasons imo: - An honest agent that steers an expert into adopting its defaults isn't misbehaving, then it falls outside the frame for funders. - Checking an individual output before someone acts on it has no established name in the grant vocabulary. - Funders focused on existential risk filter for direct paths to catastrophe. - The disciplines that know this problem best, human factors, cognitive psychology, etc. rarely apply to AI safety funders, and fieldbuilding mostly trains machine learning researchers. - It looks like a product problem (i.e. better review interfaces and audit logs), but it's not. There are plenty of nonprofits to be built here, that would let us see, measure and price what happens when a person accepts an agent’s work.Christian Catalini @ccatalini6/ Verification infrastructure is how we counter all this. So is "attention ergonomics" (see @sschillace's piece, H/T @livjenkss). 🚨 We can’t safely automate what we can’t verify, and we can’t verify what no human is paying attention to.010140- Posts @ccatalini6/ Verification infrastructure is how we counter all this. So is "attention ergonomics" (see @sschillace's piece, H/T @livjenkss). 🚨 We can’t safely automate what we can’t verify, and we can’t verify what no human is paying attention to.
215663 - Christian CataliniRT @ccatalini: 4/ 🚨 The incentives make this particularly dangerous. Productivity gains are immediate and visible. The costs of weakened oversight are often delayed, difficult to measure, or borne by someone else. Under-verification is privately rational for both individuals and firms!030747
- Christian CataliniRT @ccatalini: 1/ In February, we argued that the cost of execution would fall much faster than the cost of verification. As agents scale, human attention and expertise become the binding constraint. @AnthropicAI just put experimental evidence behind our warning...Alex Tamkin @AlexTamkinNew Anthropic research: How AI agents shape experts' judgment calls When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.0504.2K
- Posts @ccatalini7/ Preserving context, exposing consequential assumptions, and making alternatives easy to evaluate are how we expand verification capacity alongside agentic execution.Christian Catalini @ccatalini1/ Some Simple Economics of AGI—🔥🧵 Right now, there is a low-grade panic running through the economy. Everyone is asking the same anxious question: what exactly is AI going to automate, and what will be left for us?004506
- 1/ Some Simple Economics of AGI—🔥🧵 Right now, there is a low-grade panic running through the economy. Everyone is asking the same anxious question: what exactly is AI going to automate, and what will be left for us?
1543882.1K673K - Christian CataliniRT @TokenizedPod: Fintech is now ~4% of global financial services revenue and growing 4x faster than incumbents. So what should a bank do about it? Founder at MIT Cryptoeconomics Lab, @ccatalini on Tokenized recently said that banks should disrupt themselves before Fintech does. “I think if I was leading a global bank, I would just embrace it and rather disrupt myself than waiting to be disrupted.” “Right now, customers will want more choice. They will want better products and services. You can be ahead of that game.” “If you look at the music labels, the ones that embraced streaming and understood that the economics of distribution of music were going to zero, and they had to win in a different game.” “The blockbuster late fees in the financial services industry are going away, and we should play by new rules.” 🎙 Listen to the latest episode on Tokenizedpod(dot)comJevgenijs Kazanins @jevgenijsBCG: "Global fintech revenues surpassed half a trillion dollars in 2025, growing 22% year over year. That also means fintechs outgrew traditional financial services four-fold and now account for approximately 4% of total global financial services revenue".050169
BCG: "Global fintech revenues surpassed half a trillion dollars in 2025, growing 22% year over year. That also means fintechs outgrew traditional financial services four-fold and now account for approximately 4% of total global financial services revenue".
121811215.4K- Christian CataliniRT @ccatalini: 1/ Somewhere in your company today, an expert rejected a perfectly plausible answer. A controller caught a revenue figure that counted deferred bookings. A security engineer blocked a release that passed every test but would have handed an AI agent write access to the production database.020016.6K
- Christian CataliniRT @ccatalini: 🚨 You can't safely automate what you can't reliably measure and verify. Verification means checking that the agent did what you intended. As execution gets cheaper, proving the task was done right becomes the bottleneck to safe scaling.0101.3K
- Every industry gets to play prisoner’s dilemma. One company blinks. Everyone gets automated.Brendan (can/do) @BrendanFoodyWe’ll pay you up to $100K for every company you refer that licenses its data to @Mercor. We’ve partnered with > 500 companies so far. If you know a company with 100+ employees and data in code, gaming, manufacturing, healthcare or professional services, refer them at142756.9K
We’ll pay you up to $100K for every company you refer that licenses its data to @Mercor. We’ve partnered with > 500 companies so far. If you know a company with 100+ employees and data in code, gaming, manufacturing, healthcare or professional services, refer them at93561.5K548K- Skeptical of a future with many different intelligences? This runs on a Mac. 👀jack @jackWe believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.82262.5K
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.4.2K7.1K69.1K13.6M- The closed, excludable frontier that gets IPO investors excited keeps showing up as a free download…Kimi.ai @Kimi_MoonshotIntroducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: 🔗 Tech blog:62232.1K
Kimi.ai@Kimi_MoonshotIntroducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: 🔗 Tech blog:
1.3K3.2K30.2K4.1M- The marginal cost of trying one more attack has been falling for years. Attackers only need one. Mythos was not a discontinuity. If anything, we now have better tooling (open weights included!) to replace attention-constrained humans with machines and close the bugs we inevitably leave behind. H/T @alighodsi for surfacing the long-term trend.
42231.8K - Christian CataliniRT @sytaylor: Tala is a company you are not paying enough attention to Unsecured onchain credit. Embedded in wallets. $10bn lent. Exceptional risk adjusted results in "high risk" marketsTala @talamobileTala is partnering with @eldoradoio to bring embedded credit into its wallet, launching first in Bolivia. Uncollateralized consumer credit is hard to come by in Latin America. Once live, customers can get access in seconds in USDT on @tempo. Learn more:080176
Tala@talamobileTala is partnering with @eldoradoio to bring embedded credit into its wallet, launching first in Bolivia. Uncollateralized consumer credit is hard to come by in Latin America. Once live, customers can get access in seconds in USDT on @tempo. Learn more:7107633.8K- Christian CataliniRT @ccatalini: 5/ Fix two is the superpower. The models that ate junior tasks also collapsed the cost of learning and experimentation. Explore more fields. Test more ideas. Discover what you’re unusually good at, then use AI to do more with it. Superpowers, if you deliberately use them.a16z Frontier @a16zfrontier“You’ve just been told you have superpowers.” AI agents are starting to feel like coworkers. The cost of automation is collapsing. So what happens to jobs? To startups? To crypto? We break down the new paper “Some Simple Economics of AGI” with coauthor @ccatalini, including: • Why verification becomes the bottleneck • Practical advice for early- and late-career builders • The risk of a hollow economy, and the path to an augmented one • Why crypto may be essential infrastructure for identity, provenance, and trust In conversation with @eddylazzarin and @rhackett. Full episode ↓0201.4K
a16z Frontier@a16zfrontier“You’ve just been told you have superpowers.” AI agents are starting to feel like coworkers. The cost of automation is collapsing. So what happens to jobs? To startups? To crypto? We break down the new paper “Some Simple Economics of AGI” with coauthor @ccatalini, including: • Why verification becomes the bottleneck • Practical advice for early- and late-career builders • The risk of a hollow economy, and the path to an augmented one • Why crypto may be essential infrastructure for identity, provenance, and trust In conversation with @eddylazzarin and @rhackett. Full episode ↓4441468208K
