Has AI gotten any better at stock picking?

The question of whether artificial intelligence has meaningfully improved stock-picking ability is increasingly relevant as AI-driven tools become more accessible to both institutional and retail investors. Yet the available evidence remains sobering: beating the market is difficult, and AI does not change that fundamental reality. While machine-learning models can process vast datasets, identify patterns, and execute trades faster than any human, they are also subject to the same constraints — crowded trades, unforeseen economic shifts, and the basic efficiency of markets that already price in widely available information.

The market relevance here is significant. If AI models genuinely offered a durable edge, we would expect a persistent flow of outperformance from funds that rely on them. Instead, the record has been mixed, with many AI-driven strategies struggling to justify their fees after costs. What investors should watch is not headline claims about AI "beating" humans but rather long-term, risk-adjusted performance across full market cycles. A model that shines in a bull market may behave very differently during a downturn, and that distinction is where the real measure of progress will be found.