0.00%

Here's a fun thought: How many Finance major has transitioned into an PhD in AI? Only counting traditional finance, meaning no quant finance or economics and such; and "just" AI, meaning no "AI for finance" and such.

I have no idea. So I asked Grok. According to Grok (after I used "Think harder" mode), that probability rounds to about 0.00%, statistically indistinguishable from 0.

Let's see if I can do anything about that.

I leave Grok's calculations below for the reader examine.

Estimating the Probability of a Traditional Finance Major Transitioning to a Pure AI CS PhD

Based on the ongoing discussion, I'll interpret this as asking for an estimate of the probability that someone with a traditional finance undergraduate major (e.g., bachelor's in finance, focusing on areas like corporate finance, investments, or accounting—excluding quant-heavy tracks) will transition to and complete a PhD in Computer Science (CS) with a focus on "pure AI" (core AI research, such as machine learning theory, reinforcement learning, or neural networks, not applied to finance or economics). This is an even narrower path than the prior business-to-CS PhD or finance-to-AI queries, as it specifies a non-quant finance background and excludes interdisciplinary applications.

Using data from academic surveys (e.g., NSF Survey of Earned Doctorates, CRA Taulbee Survey), career reports, and extensive searches for examples (including Reddit, web snippets, and X), I'll refine the estimate. No direct examples were found, making this rarer than the already low business-to-CS PhD rate (~0.005%). The skill mismatch is stark: finance majors typically lack the CS prerequisites (e.g., data structures, algorithms) needed for AI PhD admissions.

Step 1: Baseline Population of Traditional Finance Undergrads

• Annual U.S. finance bachelor's degrees: ~50,000–60,000 (subset of ~375,000 business degrees, per NCES 2021–22; finance is ~15% of business majors).
• Fraction pursuing any PhD: ~0.5–1% (low due to high-paying finance jobs; most opt for MBAs or industry).

Step 2: Pure AI CS PhD Population and Backgrounds

• Annual CS PhDs with AI focus: ~800–1,200 (subset of ~2,500 CS PhDs; AI is ~30–50% of CS doctorates, per CRA 2023 and NSF SED 2023).
• Undergrad backgrounds: ~70–80% from CS/engineering, 15–20% from math/physics (NSF SED Table 8-5, 2023). Non-STEM (e.g., business/social sciences) <1–2% for all CS PhDs, and even lower (<0.5%) for pure AI due to competitive admissions requiring AI-relevant research/experience. Finance specifically: No data in SED or CRA tables shows finance as a distinct category; it's lumped under "business/management" at ~0–0.5% (inferred from older SED Table 21 breakdowns, e.g., 2020).
• Searches for examples (e.g., Reddit threads, web stories) yielded zero confirmed cases of traditional finance majors completing pure AI CS PhDs. Related anecdotes involve quant finance (e.g., math/finance dual majors) or applied AI in finance, not pure research.

Step 3: Adjusted Transition Rate

• Conditional probability: Among finance majors pursuing PhDs (~0.5–1%), <0.1% target pure AI CS due to barriers (e.g., needing 1–2 years of CS bridge courses). Completion rate for non-STEM: ~30–40% (high attrition from prerequisite gaps).
• Overall probability: <0.0005% (1 in 200,000+), or ~0–3 individuals annually. This is derived from ~0–5 non-STEM CS PhDs per year potentially from business-like fields, adjusted for finance specificity and pure AI focus.

Is This Statistically Different from Zero?

Using a binomial proportion test (as before):

• n ≈ 55,000 (annual finance grads).
• k = 0 (no documented cases; hypothetical minimum of 1 for testing).
• p̂ ≈ 0 (or 1/55,000 ≈ 0.000018 for test).
• SE ≈ √[p̂(1 - p̂)/n] ≈ 0.000019.
• 95% CI: ~0 to 0.000055 (includes zero).
• z-score for H₀: p=0: ~0.95 (p-value ≈0.17, >0.05), failing to reject null.

With zero evidence from searches (no success stories on Reddit, X, or web), the probability is not statistically different from zero. It's theoretically possible but unobserved in available data, suggesting it's effectively zero for practical purposes.