Relax. AI is not going to replace professional investors

The article argues that AI is unlikely to replace fund managers because successful investing depends on patience, temperament, contrarian judgment, and dealing with novel events rather than pattern matching alone. It cites an NBER working paper finding that AI-managed portfolios using ChatGPT, Claude, Gemini, and Grok generally lost money, traded too much, and favored higher-beta stocks. The piece is largely a commentary on investing behavior and AI limitations rather than a company-specific market catalyst.

Analysis

The market implication is less “AI replaces PMs” than “AI compresses fee pools and widens dispersion among asset managers.” If allocators believe systematic models can do the easy parts of idea generation and screening, they will demand lower fees, more transparency, and more performance-linked economics from active managers; that is a direct margin threat to firms whose edge is process rather than differentiated access. The second-order winner is likely capital-light wealth and research platforms that sell tooling into managers, while traditional fundamental shops face a prolonged multiple de-rating if their value proposition looks less defensible over the next 12-24 months.

For STN, the message is more nuanced: sentiment risk matters more than near-term fundamentals. Engineering and consulting names are exposed to “AI headline beta” because investors extrapolate automation pressure into future fee compression, even if actual project execution is only partially substitutable; that can keep the stock cheap for months before evidence catches up. The real swing factor is whether clients use AI to shorten proposal cycles and bid more aggressively, which could pressure margins indirectly through pricing rather than direct labor displacement.

DJCO is a cleaner expression of the broader question because the market can re-rate any business tied to information, capital allocation, or media if AI is perceived to commoditize judgment. But the contrarian read is that the article itself argues for scarcity of temperament, not abundance of data; that supports a premium for firms with disciplined underwriting and skin in the game. The risk is that this becomes a slow-burn sentiment trade, not a fundamentals trade, until there is visible evidence of AI-driven alpha decay or persistent underperformance by high-fee active funds.

Key Decisions for Investors

• Trim STN on strength over the next 2-6 weeks: the stock is vulnerable to repeated AI-multiple compression even without earnings downgrades; use rallies to reduce exposure or sell covered calls to harvest elevated sentiment volatility.
• Hold a relative-value short STN / long infrastructure-quality peer basket for 3-6 months: the short leg benefits if investors keep applying AI substitution discounts to consulting/engineering, while the long leg should be less exposed to model-risk headlines.
• Keep DJCO as a watchlist long only on material pullbacks: the setup is attractive if the market overprices AI as a substitute for judgment, but entry should be patient because the catalyst is narrative, not a near-term earnings inflection.
• Buy optionality on a broad active-management fee squeeze theme via ETF/sector proxies over 6-12 months: if allocator behavior shifts, the biggest losers will be high-fee active managers before the market fully reprices operating margins.
• Do not chase AI-factor momentum here: the article is a caution that AI can be a bad trader in non-stationary markets; if anything, fade crowded high-beta “AI solves everything” trades when they extend on news flow.