Ken Griffin: AI agents are now completing PhD‑level finance tasks

Ken Griffin’s latest warning about artificial intelligence was less about chatbots than about the research pipeline inside elite finance. The Citadel founder said AI agents are now doing work that his firm previously assigned to people with master’s and PhDs in finance over weeks or months, compressing that cycle into hours or days. Benzinga and Crypto Briefing both reported the remarks on May 18, citing comments Griffin made at the Stanford Leadership Forum. (benzinga.com) Griffin said the pace of change left him “fairly depressed,” according to follow-up accounts of the event. (cryptobriefing.com) ### What exactly did Griffin say was changing inside finance? Ken Griffin said AI agents are now handling “extraordinarily high-skilled jobs” in finance that once required advanced academic training and much longer timelines. (benzinga.com) In one widely cited version of the remarks, he said work usually done by people with master’s and PhDs in finance over “weeks or months” is now being done by AI agents over “hours or days.” The timeline compression is the key fact in his description. Griffin was not talking about routine clerical work. He was describing research and analysis tasks that, by his account, had previously justified teams of highly trained specialists. (benzinga.com) ### Why did he say the shift left him “fairly depressed”? Griffin said he “went home one Friday, actually fairly depressed” after seeing how far the tools had advanced, according to accounts published on May 17 and May 18. Business-focused summaries of the event said he had previously been skeptical of AI and now viewed it as “profoundly more powerful” than he had expected. (cryptobriefing.com) That reaction matters because Griffin has not generally been grouped with the loudest AI evangelists. The force of the comment came from a hedge fund founder describing a change in his own operating environment, not from a software vendor marketing a product. (benzinga.com) ### Which jobs was he talking about — and which ones was he not? The remarks pointed at high-end analytical work, not just entry-level drafting. One version of the comments said these were “not mid-tier white-collar jobs” but “extraordinarily high-skilled jobs” being automated by agentic AI. That distinction is important for how firms may talk about AI adoption. (letsdatascience.com) Griffin’s example suggests the first visible impact may be on research workflows: gathering material, synthesizing prior work, drafting code, running first-pass analysis and producing preliminary outputs faster than a human team could on its own. That is an inference from the reported remarks, not a direct quote from Griffin. (benzinga.com) ### What does this change inside a hedge fund or trading firm? Citadel was the named firm in the reports, but the comments travel beyond one company because they describe a workflow shift. If AI can produce a first pass on complex finance work in hours or days, the human role moves toward setting the question, checking the assumptions and deciding whether the output is usable. (josephwdr.wordpress.com) That framing is an inference based on Griffin’s description of the work being accelerated. The immediate pressure falls on the middle of the research process. A faster draft is only valuable if someone can verify the data, challenge the model and connect the result to an actual investment or trading decision. Griffin’s remarks did not lay out a hiring plan, but they did describe a much shorter path from question to draft answer. (benzinga.com) ### Where did the comments surface, and what comes next? The Stanford Leadership Forum was the setting cited in multiple follow-up reports published on May 17 and May 18. Benzinga reported Griffin’s comments on May 18, and Crypto Briefing published a separate account the same day. (benzinga.com) Citadel has not, in the reports reviewed here, published a detailed public breakdown of which internal finance tasks are now being done by AI agents and which still require full human teams. The next concrete step is likely to come through future public appearances by Griffin or hiring signals from firms such as Citadel that show whether these compressed workflows are being formalized across research teams. (benzinga.com)