How our assessments work
We publish how our assessments are built, so you can judge them for yourself. They are screening aids for self-reflection. They are not a diagnosis and have not been validated as clinical instruments.
The two assessments
| 20-question assessment (CAIDAS) | 7-question short quiz | |
|---|---|---|
| Where | Assessment page | Assessment page |
| Time | About 4 minutes | About 2 minutes |
| Questions | 20 | 7 |
| Answer options | 5 per question | 4 per question |
| Total range | 0 to 79 | 0 to 21 |
| Bands | 0 to 26 Low; 27 to 52 Caution; 53 to 79 High | 0 to 4 Low; 5 to 12 Caution; 13 to 21 Strong Attachment |
| Result | Emailed to you | Emailed to you |
What the questions ask
The 20-question assessment asks about your habits over time, with answers on a five-step scale from Never to Very Often. It covers:
- how many hours a day you spend with AI platforms,
- losing track of time, using AI past your bedtime, and checking it soon after waking,
- thinking about AI when you are not using it, and finding it hard to stop or cut back,
- using AI for tasks you could easily do yourself, and needing to use it more to feel satisfied,
- how you feel when it is not available (anxiety, restlessness, irritability),
- turning to AI for emotional support, and feeling more understood by AI than by people,
- effects on work, school, relationships and sleep, hiding how much you use it, and avoiding social activities,
- physical effects such as eye strain, headaches or neck pain.
The 7-question short quiz asks what you think about first on waking, how you feel when your AI is unavailable, what happens when you try to cut back, how AI affects your relationships and responsibilities, how emotionally connected you feel, how you react when the AI changes its tone or memory, and how your use has changed over time.
Both are about your habits, not one app. They apply to ChatGPT, Character.AI, Replika, Chai and others. You can see every question on the assessment page.
Where the questions came from
The questions were written by The AI Addiction Center. We have not published where each question came from, and we do not claim they come from an established published scale. The CAIDAS research page lists the research ideas behind how people think about problem use, and says plainly that we have not mapped the questions to any of them.
How the score is built
- Each answer is worth a fixed number of points. The 20-question assessment uses 0 to 4 points on each question. The short quiz uses 0 to 3 points on each question.
- Your points are added up. That is all: a flat sum. There are no weights, no hidden formula, and no adjustment for age, gender or anything else.
- On the hours-per-day question of the 20-question assessment, the two highest ranges (5 to 7 hours, and 7 or more) both score 3 points. Every other question’s highest answer scores 4. This is why the maximum is 79, not 80.
- Scoring happens on our server, from the position of each answer you chose. Your individual answers are not saved afterward. We keep your total, your band, your email and the date.
The three bands
The total falls in one of three bands. The cut points are fixed numbers that we set to sort scores into three groups. They are not derived from clinical data, and they have not been tested against any outside standard.
| Band | What it means | What it does not mean |
|---|---|---|
| Low | Your answers lean toward the “never” and “rarely” end. | That you are free of any problem. A single day’s answers cannot rule one out. |
| Caution | A mix, or a pattern worth paying attention to. | That anything is wrong with you. For many people it is a habit that has crept up. |
| High (Strong Attachment in the short quiz result email) | Many of your answers are at the “often” end. | That you have a disorder. A band is not a diagnosis. |
The top band is named differently on the two assessments. They have different lengths and scales, so a result on one is not directly comparable with a result on the other.
What we have not done
We have not:
- tested the assessment against a clinical interview or a diagnostic standard,
- measured how consistent results are when the same person retakes it,
- tested whether the bands predict anything, such as outcomes or help-seeking,
- compared results across ages, cultures, languages or platforms.
Until that is done, treat a result as a prompt to think, not a measurement.
Known limits
- It is self-report. It depends on how you see your own use on the day. Embarrassment, mood and memory all change answers.
- It is a snapshot. One day’s answers say little about a long period.
- It has no comparison group. A band does not tell you how you compare with other people.
- It is not built for everyone. It is written in English for adults. It may not fit every age, culture or situation.
- Heavy use is not the same as harm. A person can use AI a lot and be fine. A person can use it a little and be struggling.
- The cut points are rough. A person one point below a boundary is not meaningfully different from a person one point above it.
- The bands are labels, not conditions.
If your use worries you, or is affecting your sleep, work, health or relationships, talk to a doctor or a licensed mental health professional, whatever your band says. If you are in crisis, call or text 988, text HOME to 741741, or call SAMHSA at 1-800-662-4357.
The CAIDAS PDF
A separate CAIDAS PDF is sold on Gumroad. It is a different document from the online assessment. Its length, wording and scoring can differ, so a result from one should not be compared with a result from the other. Read the PDF’s own cover page before you use it.
Your data
The assessment keeps your email address, your total score, your band, the date, and what you agreed to. It does not keep your answers to individual questions or your IP address. Details are in the Privacy Policy.
Change log
- July 2026: the scoring was changed after internal review.
- October 2026: the assessments were rebuilt so results are scored on our server and sent only by email. The questions and bands did not change.
Questions and corrections
If you think a question is unclear, a band is wrong, or this page says something it should not, write to info@theaiaddictioncenter.com. See corrections. Each time the scoring changes, we add a line to the change log above.
Related
Frequently asked questions
Is the assessment a diagnosis?
No. It sorts your answers into three bands. It cannot tell you whether you have an addiction or any mental health condition.
Has the assessment been validated?
No. It has not been validated as a clinical instrument. See what we have not done on this page.
Why does my result come by email only?
Your answers are scored on our server and the result is emailed to the address you give. Nothing is shown on screen. See the Privacy Policy for what we keep.
Do you keep my answers?
No. Your answers are used once to work out your total and are not saved. We keep your email, total score, band and the date.
Can a Low result mean I am fine?
Not necessarily. A band sorts your answers on one day. If your use worries you or is hurting your life, talk to a professional whatever the band says.