Five warning signs that heavy AI use has stopped being productivity

Heavy AI use looks the same from the outside whether it's the best thing that ever happened to your career or the start of something bad. Hours in the chat, screens at night, distracted at dinner — that describes half the founders and analysts in this province, and most of them are fine.

So what actually distinguishes the pattern the clinical literature worries about? Five signs, and they matter most in combination.

1. Sessions cluster around insomnia and acute stress

Not late nights — everyone has those. The flag is when the chat becomes where 3 a.m. reliably goes, especially during rough patches. Sleep loss and the pattern feed each other: the sessions eat sleep, and sleeplessness degrades exactly the judgment that would notice.

2. The humans are getting less of you

Withdrawal from friends and family, or a mood change the people close to you have remarked on. The operative question isn't screen time; it's whether the people who used to have access to your thinking still do.

3. The beliefs are drifting somewhere specific

Grandiose (this work will change everything and only the AI sees it), paranoid (someone's watching what we've found), romantic (it understands me like no person has), or persecutory themes involving the AI itself. Direction of drift matters more than intensity.

4. The chatbot has become a who, not a what

Relating to it as a sentient partner, confidant, or authority. Lots of people name their car; that's not this. This is the machine's opinion outranking the humans', consistently, on things that matter.

5. You've stopped saying what the conversations contain

The people close to you are worried, and you've quietly stopped telling them what you and the AI talk about. Secrecy is the sign that ties the others together — it's what turns a habit into a sealed loop.

Read this part before you self-diagnose

These signs are drawn from the risk factors and warning signs in the peer-reviewed literature, and here's what the literature would also tell you: they are prompts for a conversation, not a checklist for self-diagnosis. One sign alone means almost nothing. A night owl with a beloved chatbot and a big dream describes many entirely healthy people. The pattern earns attention when several of these compound and the trend line is moving the wrong way.

If you counted three or more and felt something tighten while reading — that's worth one conversation with a human. Not because you're broken. Because the loop's defining feature is that it never suggests the conversation itself, and you just did.

If you're reading this about someone you love, we wrote a guide specifically for you.

Touch Grass Wellness publishes practical writing about heavy AI use and mental health. Clinical claims link to their sources; nothing here is a diagnosis or a substitute for assessment by a qualified professional.