Editorial illustration for Tech CEOs urged to use AI heavily to gauge limits, says Levie
Tech CEOs urged to use AI heavily to gauge limits, says...
Tech CEOs are developing a strange new mental illness. It's called AI psychosis, and Allan Levie, the CEO of DocuSign, thinks he's identified the cause. The boss's chair offers a dangerously clean view of a messy process.
They see a prototype, a generated contract, a neat demo. They don't see the grunt work that comes after.
CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI,” Levie wrote on X. CEOs “play with AI,” develop a prototype, or generate a contract, to use Levie’s examples, and then make the leap to believing agents can do the work.
The disease has a predictable symptom. A leader watches a machine perform a trick and concludes it can do the whole job. This isn't optimism.
It's a failure of imagination about what work actually is. Real work is adaptation, context, exception handling. It's the dull, stubborn friction between a neat output and a useful outcome.
Levie's prescription is simple. CEOs need to get infected by that friction. They should use the tools themselves, deeply and badly, to feel where they break.
You can't outsource your understanding of a technology's limits. That knowledge is earned in the trenches, not observed from the penthouse. Stop delegating your curiosity.
Go break something.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives OpenRouter valuation jumps to USD 1.3 B as AI gateway gains enterprise traction
From Archives Deepseek makes 75% discount permanent, output tokens priced over 34× below GPT‑5.5
Common Questions Answered
What is AI psychosis according to Allan Levie?
AI psychosis is a condition affecting tech CEOs where they develop an unrealistic view of AI capabilities after seeing polished prototypes and demos. Levie argues that leaders in the boss's chair see only the neat outputs without understanding the complex grunt work and messy processes that actually make AI implementations work in real-world scenarios.
Why do tech CEOs develop an incomplete understanding of AI capabilities?
CEOs typically see only the final polished demonstrations and generated outputs, which creates a dangerously clean view of what is actually a messy process. They don't witness the extensive adaptation, context handling, and exception management that occurs behind the scenes, leading them to conclude that AI can handle entire jobs when shown a single successful trick.
What is Levie's prescription for CEOs to better understand AI limitations?
Levie recommends that CEOs should use AI tools themselves, deeply and badly, to personally experience where the technology breaks down and encounters friction. By directly engaging with these tools rather than just observing demonstrations, leaders can develop a more realistic understanding of the actual work involved in implementing AI solutions and the gap between neat outputs and useful outcomes.
How does Levie define real work in the context of AI implementation?
According to Levie, real work is not just the neat output that AI produces, but rather the adaptation, context handling, and exception management that must occur afterward. Real work encompasses the dull, stubborn friction between what a machine generates and what actually becomes a useful outcome for the organization.
Further Reading
- CEOs Must Master AI: Why, How, and Its Connection to the Bottom Line — P-Tech Partners
- Only 25% of workers are using AI. Here's how tech leaders are ... — IBM Think
- CEOs think AI use is mandatory — but employees don't agree, survey says — HR Dive
- Tech CEO's warning for American workers: AI about to wipe out millions of jobs — Axios / YouTube
- CEOs are wasting millions on AI. Here's why. — YouTube
