AI Skills Self-Assessment
Answer 20 quick questions about how you actually use AI at work. You'll get a five-dimension radar chart, your weakest areas, and exactly what to practice next — all in your browser, no sign-up.
How it works
- Answer 20 questions. Each takes about ten seconds — pick the option that matches your real behavior, not your ideal behavior.
- See your radar. We score five dimensions: AI literacy, tool proficiency, prompting, data fluency and workflow automation.
- Read your gap list. Every dimension below 60% gets three concrete improvement tips plus tools to practice with.
Yes. Every question and every result on this page is free, unlimited, and requires no account, email or sign-up. You can retake it as often as you like.
No. The whole assessment runs in your own browser. Your answers never leave this page, nothing is sent to a server, and closing the tab erases everything.
AI literacy (knowing what AI can and cannot do), tool proficiency (daily hands-on use), prompting (writing effective instructions), data fluency (working with numbers and charts) and workflow automation (removing repetitive work). Each dimension uses four real-work scenarios.
Each answer is worth 1 to 5 points. Your score per dimension is out of 20, and your total is out of 100. Any dimension below 60% (12 of 20) is flagged as a weakness with tailored tips.
Most people finish in 3 to 5 minutes. If you only have one minute, close the tab and come back — nothing is saved on a server, so just retake it later.
Who this tool is for
This assessment suits people whose job title does not include the word engineer but whose week now includes AI tools anyway. Administrative assistants, HR coordinators, marketing executives, operations staff, support agents, account managers and finance clerks all answer the same twenty scenarios. If you have been using a chat tool for a few months and cannot tell whether you are ahead or behind, the five scores give you a starting point. Team leads also use it before choosing a training budget, because the weak-spot list names the dimension to fund rather than the topic that happens to be trendy.
It is not a certificate and not a hiring test. Twenty self-reported answers cannot measure your intelligence, your value to an employer or your future. They record habits you can describe today.
How to read your result
Each of the five dimensions holds four questions, and every answer scores 1 to 5, so a dimension lands between 4 and 20. The percentage beside each row is simply that score divided by 20. The total out of 100 is the sum of the five. A 20 means you chose the top option four times in one area; it does not mean expert, and 12 is only this page's 60% line.
Read the shape of the radar before the total. Five scores packed between 13 and 16 produce a round shape and a healthy total, but also a plateau: nothing is weak enough to trigger advice. A single deep dent tells you more. Any dimension below 12 is repeated in the weak-spot block with three practice tips and a short label. If no dimension falls below 12, you get a notice instead, and the useful move is to keep stretching whichever score is lowest.
The total is a rough summary, not a grade. Two people can both score 61 with opposite profiles: one strong in literacy and weak in automation, the other the reverse. Their next steps should differ.
A worked example
A customer support team lead at a 40-person software company answers honestly. She checks AI answers against help-centre articles almost daily, so AI literacy comes out at 17/20 (85%). She opens a chat tool every morning and uses the writing assist inside her ticket system, giving tool proficiency 15/20 (75%). Her prompts are usually one line, and she rarely saves the ones that work, so prompting is 9/20 (45%). She asks AI to summarise survey comments but has never checked a chart it produced, so data fluency is 12/20 (60%). Templates exist, but nobody built them with AI, so workflow automation is 8/20 (40%). Total: 61/100.
Only two rows sit below the 60% line, so the result page shows two weak-spot blocks. The prompting block suggests adding role, audience, format and tone to her next request, sending exactly one follow-up instead of retyping, and starting a note of prompts she actually used. The automation block suggests listing her three most repetitive deliverables, building a reusable template for the worst one, and asking AI for a first draft of a task she always starts from scratch. At the bottom, two buttons lead to the tool matcher and the daily learning plan. Nothing on the page tells her she is behind; it tells her where the next hour is best spent.
Common mistakes
Answering for the colleague you would like to be is the most common one. The scale asks about frequency, so sometimes should mean some days last month, not I know it exists. A second mistake is scoring every question in the middle to avoid deciding; four 3s in a dimension produce 12/20, which lands exactly on the threshold and hides the real pattern.
People also read the total as a performance rating and feel either relieved or discouraged by a number that only describes current habits. Retaking until a preferred total appears defeats the purpose; a retake is useful after six to eight weeks of practice, when the answers should genuinely have changed. Finally, do not spend the next month on your strongest dimension. The weak-spot block exists because the lowest row usually offers the cheapest improvement.