A warning has emerged that agent-based artificial intelligence (AI) is rapidly replacing even PhD-level, highly skilled work in the financial sector.

BeInCrypto, a blockchain outlet, reported on May 18 that Citadel CEO Ken Griffin (켄 그리핀) said in a Stanford University talk that work that used to take master’s and PhD staff weeks to months now takes just hours or days.

The remarks drew more attention after Elon Musk shared a related video clip. Griffin has long been known as skeptical about AI, but said Citadel’s internal operations have changed as AI tools have strengthened sharply over the past 9 months. He said the shift is targeting roles requiring advanced expertise, not general office jobs.

Griffin told Stanford students, "This does not end as a threat to ordinary office jobs," and added, "Agent-based AI is automating areas that require expertise." He said he first felt the scale of AI’s impact after seeing the change firsthand inside his company. He also said he felt quite depressed on the way home at the time.

Citadel has hired large numbers of research staff from top-tier programs in mathematics and physics. A CEO of such a company publicly saying AI has begun directly replacing PhD-level work is being read as a signal that AI has started competing head-on with the top professional talent pool.

This trend also intersects with the restructuring drive seen across the tech industry this year. Several tech companies cited AI as the backdrop to layoffs of thousands of people this year, and agent-based systems are increasing the share of cuts. Musk has also consistently said AI could eliminate most paid labor in the long run.

Still, some views say it is too early to conclude that AI has already caused large-scale job losses across the labor market. After reviewing 4 major recent studies, a16z judged that AI is not yet broadly replacing jobs across the overall economy. The analysis said the current impact is concentrated on specific task units rather than entire roles.

Reshaping businesses around AI agents is also continuing in the cryptocurrency industry. Major employers such as Coinbase and Microsoft have recently described workforce reductions as a process of shifting to smaller teams reinforced with AI. Crypto-native companies have also been building agent-based product roadmaps that directly carry out trading and payments.

A key point to watch is whether this productivity leap actually continues through year-end. In companies’ next quarterly earnings announcements, it may become clear how concretely cost savings and workforce redeployment from adopting AI are quantified. Citadel is also drawing attention over whether it will disclose how much processing capacity it has secured by handing PhD-level work to agents.

Where will AI be in 1, 2 or 3 years? pic.twitter.com/4G82dMLMZG