What AI Is Actually Doing to Hedge Fund and Bank Analyst Roles

The "AI is coming for analyst jobs" headline is everywhere. I wanted to check what the actual labor-market data says, rather than just repeat the headline, so this pulls together what Goldman's and Morgan Stanley's own 2026 studies found.

What Goldman Sachs and Morgan Stanley's own research actually found

Both banks published labor-market studies in 2026 that scored occupations by how exposed they are to AI, separating jobs that can be largely substituted by AI (their example: proofreader) from jobs that are complemented by it (their example: doctor, work that leans on judgment, accountability, and interpersonal interaction AI cannot replace).

Goldman's finding: AI exposure has genuinely moved the unemployment rate in both directions at once, a 0.16 percentage point rise in unemployment in easily-substituted occupations, versus a 0.06 point fall in unemployment in AI-augmented occupations. Morgan Stanley ran a similar analysis and reached a similar order of magnitude: AI has added at most 10 basis points to the overall unemployment rate so far. Small in aggregate, but not nothing, and clearly uneven across job types.

Where that leaves research and analyst roles specifically

Within banking itself, Goldman, Morgan Stanley, and JPMorgan have all said publicly that AI is augmenting analysts rather than replacing them so far. The way it shows up is banks getting more output per person, not cutting headcount outright. Two concrete, more specific data points sit underneath that:

• Some firms have reportedly trimmed their entry-level analyst class sizes by an estimated 10-20%, a real, if modest, effect on the number of junior seats available, even without wholesale replacement of existing analysts.
• Routine, well-defined tasks (pitch-book drafting, first-pass document review) are the parts genuinely being automated first. That tracks with the substitutable-vs-augmented framing above: the more standardized a task, the more exposed it is.

The honest caveat on the more dramatic claims

Some industry commentary (mostly from AI-research-tool vendors themselves) makes bigger claims. One vendor blog I found while researching this cited hedge funds using generative AI achieving "3-5% higher annualized returns" than non-adopters. I am not including that as a fact here: a company selling AI research tools has an obvious commercial interest in that framing, and I could not find it corroborated by an independent source. Worth remembering that "AI and finance" content is itself a hype-prone category, exactly the kind of gap between what is promised and what is shown that this site's own Hype vs Fundamentals module is built to flag.

My actual read

For someone starting out in this industry, the useful question is which parts of the job are shrinking and which are not: the standardized parts (pitch-book drafting, first-pass document review) are the ones going first, and the judgment, interpretation, and client-facing parts are where the value, and the entry-level seats, will concentrate.

Sources: AI's impact on the job market is starting to show up in the data — Axios, Can AI Replace Wall Street Analysts in 2026? — Impact Wealth, AI in Hedge Funds: Use Cases, Risks, and Best Practices — AlphaSense