Finance's AI race has a clear leader. ๐—”๐—ป๐—ฑ ๐—ถ๐˜'๐˜€ ๐—ป๐—ผ๐˜ ๐˜๐—ต๐—ฒ ๐—ฏ๐—ฎ๐—ป๐—ธ๐˜€. At trading firms like Jane Street and Optiver, over ๐Ÿญ ๐—ถ๐—ป ๐Ÿญ๐Ÿฌ job ads are for AI roles. At banks, about ๐Ÿฏ ๐—ถ๐—ป ๐Ÿญ๐Ÿฌ๐Ÿฌ. That's from a new ๐—˜๐—–๐—• ๐——๐—ถ๐˜€๐—ฐ๐˜‚๐˜€๐˜€๐—ถ๐—ผ๐—ป ๐—ฃ๐—ฎ๐—ฝ๐—ฒ๐—ฟ by รlvaro Cartea, Thierry Foucault and others (Oxford, HEC Paris, ECB). They review the research and add fresh data, starting from one simple idea: AI squeezes more insight out of data, with less human effort. ๐—•๐˜‚๐˜ ๐˜€๐—บ๐—ฎ๐—ฟ๐˜๐—ฒ๐—ฟ ๐—บ๐—ฎ๐—ฐ๐—ต๐—ถ๐—ป๐—ฒ๐˜€ ๐—ฑ๐—ผ๐—ป'๐˜ ๐—ฎ๐˜‚๐˜๐—ผ๐—บ๐—ฎ๐˜๐—ถ๐—ฐ๐—ฎ๐—น๐—น๐˜† ๐—บ๐—ฒ๐—ฎ๐—ป ๐˜€๐—บ๐—ฎ๐—ฟ๐˜๐—ฒ๐—ฟ ๐—บ๐—ฎ๐—ฟ๐—ธ๐—ฒ๐˜๐˜€. Firms can also use AI for edges with little social value, like racing to trade on public news or reading other traders' footprints. That can crowd out real research. ๐Ÿ“‰ In Dutch blue chips, traditional investors now do under ๐Ÿฐ๐Ÿฑ% of their trading on the live order book, down from ๐Ÿฒ๐Ÿฑ%. The authors think many are steering clear of the high tech trading firms. ๐Ÿค– Trading algorithms can even learn to go easy on each other instead of competing, ๐˜ธ๐˜ช๐˜ต๐˜ฉ๐˜ฐ๐˜ถ๐˜ต ๐˜ฃ๐˜ฆ๐˜ช๐˜ฏ๐˜จ ๐˜ต๐˜ฐ๐˜ญ๐˜ฅ ๐˜ต๐˜ฐ. ๐—ช๐—ต๐˜† ๐—ถ๐˜ ๐—บ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€: AI could make markets smarter and fairer, or tilt them toward whoever has the best data and models. The authors' verdict: ๐˜ฏ๐˜ฐ๐˜ต ๐˜ฅ๐˜ฆ๐˜ค๐˜ช๐˜ฅ๐˜ฆ๐˜ฅ ๐˜บ๐˜ฆ๐˜ต. Which way would you bet? ๐Ÿ” Caveat: it's mostly a review, and the new evidence is descriptive and largely Dutch, so it shows trends, not proof of cause. ๐Ÿ“„ ๐—”๐—ฟ๐˜๐—ถ๐—ณ๐—ถ๐—ฐ๐—ถ๐—ฎ๐—น ๐—ถ๐—ป๐˜๐—ฒ๐—น๐—น๐—ถ๐—ด๐—ฒ๐—ป๐—ฐ๐—ฒ ๐—ฎ๐—ป๐—ฑ ๐—ณ๐—ถ๐—ป๐—ฎ๐—ป๐—ฐ๐—ถ๐—ฎ๐—น ๐—บ๐—ฎ๐—ฟ๐—ธ๐—ฒ๐˜๐˜€ ๐Ÿ”— #ArtificialIntelligence #AI #MachineLearning #GenerativeAI #Finance