Our Take
Levie's critique is not anti-AI; it's pro-competence—executives need to touch the work to know if the productivity gains are real.
Why it matters
As AI layoffs accelerate and adoption accelerates in parallel, the gap between C-suite theory and ground-truth performance is widening. CEOs betting on tiny teams replacing large ones need to verify that bet.
Do this week
Engineering lead: Audit your team's AI tool adoption this week—are managers actually using Claude or ChatGPT for their workflows, or just reviewing dashboards? Bring results to your next planning meeting.
Box founder diagnoses executive distance from AI reality
Aaron Levie, founder of Box, sparked a debate this week by claiming tech CEOs are "uniquely prone to AI psychosis" because they operate too far removed from the actual work AI is meant to augment. On TechCrunch's Equity podcast, reporters Kirsten Korosec, Sean O'Kane, and Anthony Ha unpacked the comment, finding it less a rejection of AI tools than a call for accountability in how they are deployed.
Levie's argument centers on a structural problem: distance. Executives and venture capitalists funding startups are enamored with a vision of drastically smaller teams producing the same output, but many have never used the tools themselves to test whether that math holds. "If you're not really touching any of the end work, how would you know?" Ha summarized the critique during the conversation.
The comment landed amid visible user friction with AI integration. DuckDuckGo reported installs up 30% after users rejected Google's AI-heavy search overhaul. Meanwhile, Google itself made a public relations misstep when its AI search tool failed basic tasks—misspelling the company's own name when asked to count the P's. These are not hypothetical failures in a lab; they are live embarrassments in the product millions use daily.
Competence and credibility gap widens as AI adoption accelerates
Two contradictory truths are both accurate: AI tools are widely adopted and loved by users, and a significant audience actively rejects AI-first products. That polarization exposes what Levie identified—a chasm between those betting their business model on AI and those actually validating whether the bet makes sense.
The stakes are concrete. Companies are laying off workers based on AI productivity claims they have not personally verified. Korosec noted the dynamic directly: "These companies are using these tools, and it is directly affecting workers in the form of layoffs, and also the way that they work. The two truths are accurate here." A CEO who mandates a 30% workforce reduction based on AI efficiency gains but has spent zero hours debugging a model's hallucinations in production is running on faith, not data.
Levie's observation also challenges the current sales narrative around AI. Vendors and startup pitches have centered on cost reduction through smaller teams. If that pitch is being bought by leaders who have not stress-tested the claim, adoption decisions are being made on slides, not results. The gap between promise and proof is where credibility evaporates and user backlash takes root.
Use the tools yourself before you mandate them
The practitioner read is direct: Levie is not saying avoid AI. He is saying understand it. If you are a manager or executive considering headcount reductions based on AI gains, spend a week actually using those tools in your own workflows. Not a demo. Real work. Note where they save time, where they fail, and where you still need human judgment.
If you are shipping AI features or integrations, the same applies. The user backlash against Google's AI search is not because AI is bad; it is because Google pushed AI into a product category (information retrieval) where users had not asked for it and where the execution was visibly broken. Specificity and restraint matter more than breadth.
For teams, the question Levie raises is operational: Are your leaders actually using the AI tools your team has adopted, or are they reading quarterly dashboards and nodding? If the latter, you have a credibility problem. Accountability flows from proximity to the actual work.