Scanned by Gen AI tools. Everything on this page came from AI tools reading public web pages. No person or official assessor rated me. The levels are AI judgements, not test results or certificates, and the tools can be wrong. Official scores come only from the tests themselves.
what I asked
I wanted to know if this site shows a clear level of AI skill, measured against Singapore’s own scales. Perplexity and Gemini got the same question, word for word:
Read my website and its linked pages. Search for Singapore’s AI skills levels and competency scales. Look at what each level requires. Based only on the evidence on the website, what level do I show? Name the scales, map evidence to each skill, give an overall level, list the gaps, and say what would move me up one level. Be candid, not flattering.
Then Claude Opus checked both answers, and Claude Sonnet wrote up the result.
the four scales
| Scale | Who runs it | Levels |
|---|---|---|
| AI user levels | MDDI / IMDA, scored through SIT’s AI Ready Quiz | AI Aware → AI Literate → AI Fluent |
| AI learning pathways | Institute of Digital Government (IDG) | Basic Product Thinking → AI Aware 101 → AI Apply 201 → AI Build 301 |
| Skills in Generative AI | IMDA Skills Framework for ICT | 9 skills, each Basic / Intermediate / Advanced, split into an AI User tier and an AI Practitioner tier |
| AI Readiness Index (AIRI / pAIRI) | AI Singapore | L0 AI Unaware → L1 Aware → L2 Ready → L3 Competent → L4 Catalyst |
how the tools worked it out
- 1 · same question
Two tools, identical wording
Perplexity (GPT-6.1 Sol, Search mode) and Gemini (Flash).
- 2 · read the pages
Only what is written down
Each opened the home page and some linked pages: the toolkit, “Six Months of Teaching Myself AI”, “Twenty Years of Trying Tools”, the Skills Training Landscape and “Testing AI Voices”. They did not run my tools or look at how they are built.
- 3 · find the scales
Search for the official pages
Perplexity drew on 40 web sources, mostly IDG, IMDA, SIT and SkillsFuture pages. Each tool picked the scales it thought fit, so their choices differed.
- 4 · match
Evidence against level descriptions
Each scale says what a person at each level can do. IMDA’s Basic level for building AI apps means a Gen AI app that looks up information before it answers. My offline notes tool finds the five most relevant notes and answers only from them, so that counts as Basic. There is no points score. The tool judges how well the evidence fits the wording.
- 5 · check
A third model checks the first two
Claude Opus read the same pages, opened the official scale pages, and tested each claim against the site. It flagged what was wrong, made up or rated too high.
- 6 · write up
A fourth model writes the summary
Claude Sonnet combined the three answers, with no new research.
- 7 · check the checker
Back to the source
Opus said IDG has three pathways. It had only looked at IDG’s home page. The pathways page lists four starting points, so that correction stayed out of the result.
what they found
| Scale | Perplexity | Gemini | Final, after checks |
|---|---|---|---|
| AI user levels | AI Fluent | not used | AI Fluent |
| IDG pathways | Build 301 | Build 301, “firm pass” | Working at Build 301. Capstone not yet done |
| IMDA Gen AI skills | Intermediate in some, Basic in others | not used | Basic in both user skills. Practitioner skills: Basic in building AI apps and responsible AI, partial in prompt engineering and choosing models, below Basic in testing AI output |
| AI Singapore pAIRI | not used | L3 Competent (Builder) | Can’t be judged from a website. Rough guess L2–L3 |
What the site shows best
- The “Six Months” page: my notes chatbot gave confident wrong answers, so I rebuilt it smaller. I checked a viral claim against real download sizes and listed the work I haven’t finished.
- Four live tools for SMEs, each built around one real user’s job. Built with AI, though they don’t use AI when they run.
- A clear split between private material, kept on local models, and safe material sent to online tools, turned into a first-month plan for small firms.
What’s missing
- No measured result: no time saved, no before and after.
- No proof the method works for someone else. No SME or learner case yet.
- No published test of my own tools against questions I know the answers to.
where the tools got it wrong
| Claim | Tool | Check result |
|---|---|---|
| An “SF for ICT AI Application, Level 1–6” scale | Gemini | No such skill found and no link given. Probably made up |
| “No retrieval system” | Gemini | Wrong. The offline notes tool is one |
| “Deploys on Manus AI agent environments” | Gemini | Overstated. One experiment, marked “may go offline” |
| Intermediate on several IMDA skills | Perplexity | Too high. IMDA puts Intermediate in the practitioner tier, which the evidence doesn’t reach |
| IDG has three pathways | Opus | Wrong. The pathways page lists four |
| Missed my SMU and SCTP certificates | Perplexity, Gemini | Both are listed on the home page |
The tools also caught a mistake on this site. My AI Readiness Index line was out of date: it stopped at “AI Competent”, said 15 minutes and left out the personal version. I fixed it the same day.
what to take from this
- These are judgements, not scores. Ask again and the answer may move up or down a level.
- Tools see only what is written down. A skill I have but haven’t shown here counts for nothing. Hence the same gap kept coming up: no measured result.
- One answer wasn’t independent. Gemini hung on three new chats and only replied inside an older chat that already held its earlier verdict. Perplexity’s fresh answer was more reliable.
- Every layer made mistakes, including the checker. A second model caught most of them. Checking the source pages myself caught the rest.
- For a real score, take the tests: SIT’s AI Ready Quiz, AI Singapore’s pAIRI, and the IDG Build 301 capstone.
next
- 1 · numbers
One case study with numbers
Time before and after, and the wrong-answer count on 10–20 known-answer questions.
- 2 · someone else
Try the first-month plan with a small firm
One or two firms, then publish the anonymised result.
- 3 · real scores
Take the tests
SIT’s quiz and pAIRI, with the dated scores posted here.
Scales described from the public pages of IDG, MDDI, SIT, IMDA and AI Singapore on 25 Oct 2026. Not affiliated with or endorsed by any of them.