Using AI for Ten Minutes Weakens Your Persistence

Key takeaways

• Ten minutes of AI use measurably reduces persistence on difficult tasks, according to new Berkeley research.
• The risk is not laziness but a recalibrated sense of what problems warrant effort.
• This does not mean AI is bad for business, only that deployment should be deliberate.
• Preserve at least one recurring hard problem that your team solves without AI assistance.
• The strongest teams will use AI strategically and know exactly where not to use it.

The Development

Researchers at UC Berkeley recently published findings (https://vcresearch.berkeley.edu/news/using-ai-just-10-minutes-erodes-your-ability-persist-hard-things) that should give business owners pause. Their study showed that people who used AI assistance for just ten minutes demonstrated measurably less persistence when tackling difficult problems afterward. The effect appeared quickly and consistently across participants.

This is not a theoretical warning about some distant future. The study was published in October 2024 as companies across every sector are rolling out AI tools to their teams. The timing matters.

Why It Matters to a Business Owner

Persistence is not a soft skill. It is the engine that turns hard problems into solved problems. When your team hits a wall on a customer issue, a supply chain snarl, or a budget that will not close, the ability to stay with it separates businesses that adapt from those that stall.

The Berkeley research suggests that leaning on AI, even briefly, may train people to disengage when things get hard. That has immediate implications for how you deploy these tools.

If your sales team uses AI to draft emails, does that convenience erode their ability to work through a tough negotiation? If your operations manager leans on AI for scheduling, what happens to their capacity to untangle a complex logistics problem manually when the system fails? These are not hypothetical concerns. They are patterns the research predicts.

The risk is not that AI makes people lazy. The risk is that it quietly recalibrates what feels like an acceptable level of difficulty. Problems that used to warrant effort start to feel like problems to hand off. That shift happens beneath conscious awareness, which makes it harder to guard against.

What It Does NOT Mean

This research does not mean AI tools are bad for business. It does not mean you should pull them out of your workflows.

The Berkeley study measured a specific, narrow outcome: persistence on difficult tasks after brief AI use. It did not evaluate overall productivity, decision quality, or long-term learning. It did not compare teams that use AI strategically against those that do not use it at all. It did not assess whether the effect fades, compounds, or can be offset by other practices.

This is early-stage research on one dimension of human behavior. It is a signal, not a verdict. The correct response is not to reject AI. The correct response is to deploy it with your eyes open.

A Practical Next Step

If you are rolling out AI tools in your business, add one checkpoint to your process. Pick a recurring hard problem your team solves without AI. It could be a monthly reconciliation, a tricky client situation, or a process breakdown that requires manual troubleshooting. Make it a rule that this problem stays manual.

The goal is not to avoid AI. The goal is to preserve a space where persistence gets practiced. Think of it as a control group for your own operations. You get to keep the efficiency gains from AI in most of your workflows while protecting the problem-solving capacity that keeps your business resilient when tools fail or conditions change.

Our read is that the strongest teams in five years will not be the ones that use the most AI. They will be the ones that use AI deliberately and know exactly where not to use it. This research gives you a reason to draw that line now, before habits form.

Where This Is Heading

Expect more studies like this. Researchers are only beginning to measure how AI changes cognition and behavior in real-world settings. Business leaders are running this experiment live, often without baseline data.

The companies that will navigate this well are the ones treating AI adoption as a design problem, not just a procurement decision. That means asking not only what AI can do, but what it might undo. It means monitoring for second-order effects like persistence, attention, and judgment.

The Berkeley findings are one data point. But they are a data point worth acting on. If ten minutes of AI use can shift how people approach hard problems, then the cumulative effect of months or years of use is not something to guess about. It is something to manage.