Using AI makes people less likely to admit they don't know something

This is not an earnings-shift event for the named tickers; the main market implication is a second-order one: if users become systematically overconfident in model outputs, the incremental cost of AI adoption rises through higher error-checking, legal review, and workflow redesign. That favors the “trust stack” around AI more than the raw model layer — auditability, retrieval, workflow controls, and human-in-the-loop tooling should capture budget before consumer-facing assistants do. In contrast, commoditized chatbot exposure is vulnerable to a trust discount if buyers start demanding proof-of-answer, not just fluency.

The near-term catalyst is regulatory and institutional, not financial: education policy, enterprise governance, and procurement standards can move faster than product cycles, but likely over months rather than days. The biggest second-order risk is that this becomes a headline-driven anti-AI narrative that compresses multiples across the broad AI complex even though the actual revenue impact is probably tiny in the next 1–3 quarters. That would hurt high-duration beneficiaries with the richest expectations first, while firms selling compliance, security, or workflow software should see relatively better demand elasticity.

For the two named tickers, the actionable takeaway is that there is no clear edge large enough to justify a standalone position off this signal; the expected fundamental delta is effectively zero. The contrarian view is that the market may be underestimating how quickly enterprises adapt: once businesses require citations, model routing, and constrained outputs, the productivity case survives and the “hallucination” issue shifts from adoption blocker to feature-selection problem. The thesis is falsified if enterprise AI spend continues accelerating without any uptick in governance/compliance budgets, or if regulation remains symbolic and does not alter procurement behavior over the next 6–12 months.