‘We are driving in the fog’: Hundreds of economists admit they’re flying blind on AI

Over 200 economists, including Nobel laureates and the chief economists of OpenAI and Anthropic, warned that economics is “flying blind” on AI, urging action to build incentives/guardrails as AI could radically transform the economy within ~10 years. The article highlights uncertainty around AI’s employment impact and contended measurement of “AI exposure,” with real-world usage data sometimes showing less disruption than theoretical frameworks. Overall, the message is cautious—recognizing potential benefits but emphasizing near-term evidence gaps and likely disruption risk.

Analysis

This reads as an epistemic, not immediate, macro shock: when even the experts can’t agree on the exposure metric, policy and capital allocation lag the underlying labor shift. That delay is mildly constructive for the large-platform cohort because it postpones regulatory reaction and lets AI spending stay framed as optional productivity investment rather than a job-destruction story. The market should care less about headline unemployment and more about whether enterprise buyers keep funding AI pilots through existing software and cloud budgets.

The cleaner second-order loser set is labor-arbitrage and task-based intermediaries: staffing, BPO, recruiting, entry-level professional services, and some consumer lenders with young-worker exposure. If AI pressure is real, it first appears as hiring freezes, slower wage growth, and weaker renewals, then only later as layoffs; the catalyst window is 1-3 quarters for guidance language and 6-18 months for margin/credit effects. APO is not a direct beneficiary or victim, but its private-credit book would eventually feel it if lower-tier labor demand softens and consumer stress rises.

Contrarian take: the consensus may be overestimating near-term productivity gains while underestimating adoption frictions, so the first earnings impact is likely margin defense, not explosive revenue uplift. That is why platform owners with distribution and first-party data should outperform point solutions, while broad "AI exposure" baskets remain vulnerable to disappointment until usage proves monetizable. What would falsify the labor-disruption thesis is another 2-3 quarters of stable youth employment and flat wage growth in AI-exposed occupations.

Key Decisions for Investors

• Long MSFT on 5%-10% pullbacks as the highest-quality AI monetization vehicle; 6-12 month horizon, with downside only if Azure/Copilot growth fails to inflect in the next two earnings cycles.
• Pair trade: long MSFT / short IGV or a software basket if you want exposure to platform concentration versus commoditization; best entered after any broad AI software rally fades.
• No immediate trade in APO; treat it as a watch item. Consider a bearish APO/private-credit proxy only if 22-25 labor data weakens and consumer delinquency trends turn up over the next 1-2 quarters.
• Build a short list on staffing/BPO names such as MAN, KFY, or TASK, but wait for confirmation from earnings revisions or hiring data; the setup is better after the market sees the slowdown, not before.