There is no clinical diagnosis called "AI dependency" in 2026. Anyone who tells you otherwise is selling something. What there is is a recognizable working-life pattern, well-described in the Talkspace symptoms list, Psychology Today's tips for managing AI dependence, Psychreg's coverage of the cognitive-skill warning, and the Help Net Security summary of the HBR AI-fatigue report.
This post is the working person's version of the self-check. It is not a diagnostic. It is the kind of "have you noticed yourself doing this?" list a clear-headed friend would walk you through over coffee.
The pattern
The clinical-adjacent language across the sources is consistent.
Talkspace describes the experience as "working without AI feels daunting or anxiety-inducing." Psychology Today flags "reassurance-seeking" as a key warning sign. Psychreg cites the broader cognitive-skill concern: AI dependency is a warning that the practice of your own judgment is fading, not a warning that your judgment is gone.
What ties these together is a small list of things you can actually notice in yourself.
A working self-assessment
Read each, and answer honestly. None of these on its own is diagnostic. A cluster of "often" answers means the pattern probably applies to you.
- The first-reflex check. When a small thinking task comes up — drafting a single sentence, doing a quick mental calculation, summarizing a paragraph you just read — is your first reflex to open an AI assistant, even when the task would take 20 seconds unaided?
- The unaided-work anxiety check. When the AI tool you usually use is unavailable, do you feel a level of anxiety that's higher than the difficulty of the task warrants?
- The judgment-drift check. Has your confidence in your own judgment in your domain visibly dropped over the last six months? Not the quality of your output — the confidence in your judgment about your output?
- The reassurance-loop check. Do you find yourself running decisions you'd previously have made on your own through AI as a sanity-check, often, even on small calls?
- The skill-drift check. Have you noticed a real change in the quality of your unaided work — work you'd produce without any AI in the loop? Either way is interesting.
- The recovery check. After a long stretch of AI-heavy work, do you feel a kind of mental fatigue that doesn't match the difficulty of the work you actually did?
- The ownership check. When you ship a piece of work that AI helped with, do you feel meaningfully less ownership of it than you used to feel of comparable work?
If three or more of these land as "often," the pattern is plausible. If five or more land, the pattern is probably real.
What to actually do
The strongest practical advice across the sources clusters around four habits.
1. Bound AI to windows
This is the single highest-leverage move from the Psychology Today list. Pick the parts of the day where AI is in your workflow heavily, and put the rest of the day in a different mode. The reassurance loop gets harder to maintain when there's no chat window open.
2. Separate thinking from generation
You can use AI a lot and still keep your judgment in shape, as long as the thinking — what to do, what matters, what shape the work should take — happens before the generation. AI for "make this draft tighter" is fine. AI for "decide what I want to say" is where the dependency builds.
3. Recover deliberately
The HBR/BCG data is unambiguous: AI-heavy work produces real mental fatigue. Treat the recovery seriously. Step away from the screen. The fatigue is not laziness — it's load.
4. Keep one daily unmediated moment
This is the part Senwitt cares most about. A short, daily, unmediated practice window keeps the underlying judgment in regular use. It does not need to be long. It needs to be daily, and it needs to happen without AI in the loop.
That's what the Senwitt daily Set is built for. Pick 3 to 6 of the Skills — writing, math, code, memory, reading, reasoning — and complete one short Set a day. The whole thing takes about seven minutes. Nothing about it depends on AI being absent from the rest of your day.
What this self-assessment is not
It's not a diagnostic. It cannot tell you anything clinical about you. If your concern is about specific cognitive change — memory loss, sustained concentration problems, executive-function changes — that conversation is with a doctor, not with an app or a blog post.
It also cannot tell you whether AI is "right" or "wrong" for your job. That depends on your job. The point is to keep noticing — to keep the dial calibrated to "tool that helps me think" rather than slipping into "first thing I reach for whether or not it helps."
A small caveat on self-assessment
Self-assessment is useful — and it has well-documented limits. The skills you've been actively practicing tend to feel sharper than they are because the practice is recent. The skills you've been quietly delegating tend to feel intact because you haven't tested them recently. The self-check above is most accurate when you pair the answers with at least one actual unmediated task per category. Try drafting one sentence without AI, doing one mental calculation, recalling one specific thing on purpose. The contrast between "I think I can do this" and "I just did this" is where the honest signal lives.
Why a daily window is the lever that works
The four habits above — windows, separation, recovery, daily practice — are the standard published recommendations because they collectively change the relationship with AI rather than the volume. You can use AI a lot and stay calibrated, as long as those four guardrails hold. You can use AI very little and still drift if those guardrails don't hold. The volume is not the question. The structure is.
The reason Senwitt focuses specifically on the daily-practice piece is that it's the easiest piece to drop and the hardest piece to notice you've dropped. AI windows are visible — you set them or you don't. Recovery is visible — you take it or you don't. Practice is the silent one. You don't notice you've stopped doing it until the unmediated task arrives and your hands feel a little slower than they used to. The daily Set is small enough that it doesn't compete with the rest of your day, and structured enough that you don't have to spend mental energy picking what to practice.
One small word on the word "dependency"
We've used "dependency" throughout this post because that's the term the published coverage uses, and because it's the most recognizable framing of the pattern. It is worth flagging that the word is borrowed from clinical vocabulary it doesn't quite belong to. There is no AI-dependency diagnosis in the DSM. The Talkspace, Psychology Today, and HBR coverage all use "dependency" colloquially — to describe a pattern of overuse and over-trust, not a clinical condition. The framing is useful precisely because it's familiar. The accuracy bar is to keep using it descriptively, not diagnostically.
