How Much Will AI Affect Finance Jobs and Careers Over the Next 10-15 Years?
A grounded look at how AI is likely to reshape finance careers over the next decade — what's genuinely at risk, what's durable, and how to position yourself.
Nobody can predict this precisely, and anyone claiming certainty about a 10-15 year horizon is overstating their own confidence. What's more useful than a prediction is reasoning carefully from what AI tools are actually good at today, what they consistently struggle with, and how those patterns are likely to extend forward — while staying honest about the genuine uncertainty involved.
What AI is already changing, right now
The clearest, least speculative impact is on the mechanical, repetitive layer of junior finance work: first-draft formatting, pulling and organizing data, building initial comp sets, and drafting routine sections of documents. This is already happening at many firms, and it's reasonable to expect it to deepen over the next few years as the tools improve and firms build more internal workflows around them.
What this has not meaningfully touched yet is judgment: deciding which deal is worth pursuing, negotiating terms, building genuine client relationships, and exercising the contextual reasoning that separates a defensible investment thesis from a plausible-sounding one. These remain stubbornly human tasks, not because AI can't generate plausible-sounding output for them, but because the output can't be trusted without a human who understands the underlying logic well enough to catch when it's wrong.
A reasonable 2-5 year outlook
Expect continued, visible adoption of AI-assisted tools for the mechanical layers of junior work — formatting, first-draft modeling, document summarization. The junior role likely shifts incrementally from "the person who builds the model" toward "the person who builds the model well enough to know when the AI-assisted version is wrong." This is a genuinely different skill emphasis, and it rewards exactly the kind of deep, rebuild-it-yourself understanding this site's courses are built around — not surface-level tool fluency, which commoditizes fast, but the judgment to evaluate output critically.
A more uncertain 5-10 year outlook
This is where genuine uncertainty sets in. It's plausible that AI tools handle a meaningfully larger share of what junior analysts currently do by hand — comps construction, initial DCF builds, first-pass diligence summaries — which raises a real structural question the industry hasn't resolved: if the traditional "learn by doing the grunt work for two years" apprenticeship model shrinks, how do future senior bankers and investors develop the judgment that currently comes from having built hundreds of models by hand? This is a genuine open problem, not a solved one, and reasonable people in the industry disagree about how it plays out.
The skills that make someone a good senior dealmaker or investor — pattern recognition, judgment under uncertainty, knowing when a model's assumptions don't pass the smell test — are traditionally built through the repetition AI is now partially automating. If that pipeline narrows, the industry may need new, deliberate ways to build judgment that don't rely purely on junior-level repetition. This is exactly why hands-on, actually-build-it learning matters more now, not less — it's a more deliberate substitute for reps you might otherwise get on the job.
A genuinely speculative 10-15 year outlook
At this horizon, honest uncertainty is the only intellectually defensible position. Possible shapes this could take: smaller analyst classes with AI handling more of the current junior workload; a bifurcation between highly automatable transactional work and highly judgment-dependent advisory and investing work; or new finance career paths entirely, built around managing and validating AI-assisted workflows rather than executing them by hand. All of these are plausible. None are certain.
How impact likely differs by career path
- Investment banking (junior execution-heavy roles): likely the most exposed to near-term automation of mechanical tasks, though deal judgment and client relationships remain durable.
- Private equity and private credit: underwriting judgment, negotiation, and portfolio-level decision-making are harder to automate convincingly; the mechanical modeling layer is more exposed, similar to banking.
- Hedge funds: genuinely mixed — some quantitative strategies directly incorporate AI/ML as a core tool (not a threat to the role but part of it), while fundamental, judgment-driven strategies rely on exactly the kind of variant-perception thinking that's hardest to automate.
- Asset management: similar pattern to hedge funds — portfolio construction and client relationships remain durable, while research support functions are more exposed.
What this means if you're building a finance career now
- Build genuine understanding, not just tool fluency. Knowing how to prompt an AI tool is a commodity skill that will only get easier to acquire. Knowing whether its output is actually correct is not, and that only comes from having built the real thing yourself.
- Lean toward roles and skills emphasizing judgment under uncertainty over purely mechanical execution, where realistic given your starting point and constraints.
- Treat AI fluency as additive, not a substitute for the underlying technical skill — the analysts who thrive in an AI-augmented workflow will be the ones who can catch a subtly wrong assumption in an AI-generated model, which requires understanding the model cold in the first place.
This is, not coincidentally, exactly the philosophy behind every course on this site: real, working models you build yourself, not templates to fill in — the kind of understanding that doesn't commoditize.
The skill that doesn't commoditize is understanding a model well enough to know when its output is wrong. Our Valuation course builds exactly that, from the ground up.
See the Valuation course → $79