I took UD's AI IQ Test expecting a lead-capture form with a quiz painted on top. It is not that. It is fifteen true-or-false statements, split across eight industry tracks, with no payment step and no account to create, and the result is a label rather than a score report. That is a narrower product than the marketing suggests, and knowing exactly how narrow is what tells you whether it is worth your ten minutes.
This page exists because nothing else states what is actually inside these free tools before you click into them. Below is the full contents, the cost, the four score bands, the sibling tool most people miss, and an honest list of what these tests cannot tell you.
What is UD's AI IQ Test, and what is inside it?
The UD AI IQ Test is a free fifteen-question true-or-false assessment of how well you understand AI in a working context. You choose one of eight industry tracks first, and the questions are framed around that industry. It scores you into one of four bands by percentage correct and takes roughly ten minutes.
The eight tracks are catering, retail, real estate, education, medical, logistics, insurance and human resources. Choosing a track changes the scenarios in the questions, not the format or the number of them. If your work spans several sectors, pick the one where you would actually have to defend an AI recommendation to a colleague, because that is where a wrong answer costs you something.
The format matters more than it looks. True-or-false is a recognition test. It measures whether you can spot a wrong claim about AI when someone puts it in front of you, which is a real and underrated skill in a market full of confident vendor statements. It does not measure whether you can build anything.
Do you need to pay, sign up, or give a card?
No. There is no price, no plan, no free tier limit and no card. The test is open at the URL, you pick an industry and start. UD's contact details sit at the bottom of the page for anyone who wants a consultation afterwards, but nothing gates the questions themselves.
The facts, in one block
--- Cost: free. No published price, no paid tier, no HK$ figure attached to the test itself.
--- Card required: no. No payment step appears at any point.
--- Account required: no sign-up is requested to take the test or to see your band.
--- Length: 15 true-or-false statements, one industry track chosen at the start.
--- Industry tracks available: 8 (catering, retail, real estate, education, medical, logistics, insurance, HR).
--- Result format: one of four labelled bands based on percentage correct, shown immediately.
--- What happens at the end: no expiry, no trial clock, no downgrade. You can simply close the tab.
--- Follow-up channels stated on the page: WhatsApp (852) 9696 7545, phone (852) 2554 7545, and sales@ud.hk.
The absence of a payment step is the whole point of a tool like this, and it is also why you should read the next two sections before treating the result as meaningful. Free assessments are cheap to take and cheap to over-interpret.
What do the four score bands actually mean?
The test sorts you into one of four bands by percentage correct, each with a label. They are deliberately blunt, and the boundaries are published on the page before you start.
The bands
--- 0 to 25 percent, "Digital Dinosaur": the framing is that you need to understand AI immediately.
--- 26 to 50 percent, "Observer": you know a little, but there is a great deal you do not.
--- 51 to 75 percent, "Chaser": not bad, with room to improve.
--- 76 to 100 percent, "AI Pioneer": you are treated as an AI-fluent user.
If you already use AI daily, you should expect to land in the top two bands, and landing there tells you almost nothing new. The useful signal is the specific statements you got wrong. Note them before you close the tab, because there is no saved history to go back to.
Here is how to turn a band into something you can act on. Paste this into whichever assistant you use, with your own wrong answers filled in.
Try this prompt
--- CONTEXT: I scored in the [band] band on a 15-question true-or-false AI knowledge test framed around the [industry] sector. Below are the exact statements I answered incorrectly.
--- STATEMENTS: [paste each statement and whether you answered true or false]
--- TASK: For each statement, tell me (1) the correct answer, (2) the specific misconception my wrong answer implies, and (3) one concrete task in my daily work where that misconception would produce a bad decision.
--- THEN: group the misconceptions into at most three themes, and for each theme give me one 20-minute exercise I can do this week to close it. Exercises must use a tool I already have, not a tool I would need to buy.
--- CONSTRAINT: if a statement is genuinely ambiguous or depends on the model version, say so instead of picking a side.
That final constraint is doing real work. A number of common AI claims were true of one model generation and false of the next, and an assistant that is willing to say "this depends on the version" is more useful to you here than one that confidently adjudicates.
How does AI vs Human Battle differ from the IQ Test?
Most people find the IQ Test and stop there, which misses the more interesting of the two tools. AI vs Human Battle is a timed head-to-head format with twenty industry arenas, where you and an AI worker receive the same task and both have to make a judgement call within a time limit. The comparison, not the score, is the output.
The twenty arenas cover a much wider spread than the IQ Test's eight tracks, including catering food safety, retail inventory, beauty skin analysis, property matching, education diagnostics, medical triage, logistics dispatch, tax, fitness coaching, hotel concierge, legal contracts, insurance underwriting, renovation quoting, auto repair, pet care, wedding planning, HR recruitment, food delivery dispatch, construction safety and floristry.
For a practitioner, the value is the specific shape of the disagreement. When you and the AI reach different judgements on the same brief, the interesting question is not who won. It is whether the AI was working from a rule you did not know, or from a plausible-sounding generalisation that happens to be wrong in your sector. That distinction is exactly the one you have to make every day when reviewing AI output at work.
What these free tests do not do
Every honest evaluation page needs this section, and here it applies to UD's own tools.
They are Cantonese-only. Both the AI IQ Test and AI vs Human Battle are presented in Traditional Chinese with Hong Kong Cantonese phrasing. There is no English or Simplified Chinese version at the time of writing. If you work in an English-first team, you can take them yourself but you cannot circulate them.
The framing is operator-level, not tool-level. The questions are written around business adoption scenarios rather than prompting technique, model selection or workflow design. A marketer who has never touched a system prompt can score in the top band. That is a deliberate design choice for UD's audience, and it means a high score is not evidence of hands-on capability.
The bands are labels, not benchmarks. There is no published methodology, no peer comparison, no percentile, and no statement of how the questions were validated. Treat "AI Pioneer" as encouragement, not as a credential.
Nothing is saved. No account, no certificate, no result history, and no stated way to retake and track improvement over time. Screenshot your wrong answers or you lose them.
The trivia panel is uncited. The IQ Test page carries a set of AI statistics in a sidebar. None of them name a source, so do not quote those figures in a deck. Use them as conversation starters only. If you need citable numbers, our article on what the 2026 staffing numbers actually say is a better starting point.
Which one should you take, and what happens next?
Take the IQ Test if you want a ten-minute audit of your blind spots and you read Cantonese. Take AI vs Human Battle if you already know you are AI-literate and want to see where your judgement and a model's judgement diverge on a task in your own sector. Take both if you are preparing to argue for an AI budget internally, because the two together give you a diagnosis and a demonstration.
Do not take either if you are looking for a certificate, a benchmark you can put on a CV, or an English-language assessment. Those are not what these are.
After the test, the page offers a free consultation through the channels listed above. That is the actual commercial path, and it is worth naming plainly: the tests are free because UD wants the conversation that follows. Whether that conversation is useful to you depends on whether you have a real deployment question, which is a different thing from a curiosity question.
The takeaway
A free assessment is worth taking when you know exactly what it measures and exactly what it does not. UD's AI IQ Test measures whether you can spot a false claim about AI in your industry, in fifteen statements, in Cantonese, for nothing. That is a genuinely useful narrow thing, and pretending it is broader would waste your time and ours.
The part worth keeping is the list of statements you got wrong. Everything after that is a conversation about what to build, and that conversation is the one we are actually good at. We understand AI. We understand you better. With UD by your side, AI doesn't feel cold.
Reviewed by the UD AI team, Hong Kong, 31 August 2026. Product details verified against the live product pages on 31 August 2026.
From a Score to a Working System
Knowing your blind spots is the diagnosis. Turning that into a workflow your team actually runs is the build. We'll walk you through every step, from picking the right task to tool setup, prompt design and handover.
