Assessing a manager's AI capability
In one sentence: A manager's AI capability is assessed in six dimensions — operational use, critical judgment, collaboration, innovation, process and adaptability — because "uses AI" and "can be relied on with AI" are different states, and only the second one survives a bad output.
The answer first
How do you assess a manager's AI capability?
Assess six dimensions separately and never collapse them into one score: operational use, critical judgment, collaboration, innovation, process, adaptability. Each dimension is written as an observable behaviour, not as a self-rating.
- Operational use — can the manager finish real work with AI tools, at the standard the work requires.
- Critical judgment — can the manager tell which part of an AI output is wrong, invented, or cannot be trusted.
- Collaboration — can the manager run work with AI and people together, without the handover losing information.
- Innovation — can the manager open work that was not possible before, rather than speed up work that already existed.
- Process — does the organisation's mechanism carry AI output into decisions, or does it stop at the individual's screen.
- Adaptability — when the model changes under them, does the manager's way of working change with it, or does it break.
The six are not additive. A manager who scores high on operational use and low on critical judgment is a risk, not an asset: they produce more output than anyone can check. This is why the instrument reports a profile rather than a total, and why it never publishes a ranking of people.
Why not just test tool proficiency?
Because tool proficiency is the dimension that decays fastest and matters least. Every model release makes the previous year's prompt habits obsolete, and every interface change removes the advantage of muscle memory. What does not decay is judgment: knowing what to hand over, what a good output looks like, and which sentence in a confident answer is fabricated.
George Guo's test of whether AI has entered management is not tool use at all. It is whether five moves have changed: how work is assigned, how standards are set, how output is verified, whether it reaches performance review, and how the team is led. Capability assessment measures the person behind those moves.
The instrument
What does a manager receive from AIQ?
Three things, all behavioural. A six-dimension profile showing where the manager's AI work holds and where it fails. A written description of the failure pattern in each weak dimension, in the language of management rather than the language of software. And a short list of which dimensions to work on first, in the order that changes output fastest.
There is no total score and no league table. The instrument does not rank managers against each other, and it does not convert the profile into a hiring decision on its own.
What AIQ does not do
- It does not predict performance. It describes current behaviour under AI; it does not forecast next year's results.
- It does not replace an interview. For senior AI roles, it narrows the questions to ask; it does not answer them.
- It does not make a manager an AI Manager. A high profile means the capability is there. Whether AI has entered that manager's daily actions is a separate check, done against the five moves.
- It is delivered in Chinese. The instrument, the read-out and the behavioural descriptions are Chinese-language. English delivery is arranged case by case.
Is AIQ the same thing as an AI Manager?
No, and the difference matters. AIQ measures a capability: what this person can do with AI. The AI Manager category describes a state of practice: whether AI has entered how this person assigns work, reads results and leads people. A manager can score well on AIQ and still run the team exactly as before. Capability is the fuel; the five moves are whether the engine is connected.
An AI Manager, in George Guo's definition, is a manager who has brought AI into everyday managerial actions. The test is not whether they can use the tools, but whether how they assign work, read results and lead people has changed because of AI.
Evidence and its limits
What does the reference pool show?
Since 2024 the instrument has accumulated more than 900 individual manager assessments, across listed companies, growth-stage companies and state-owned enterprises, in six cities. Two findings have held across that pool.
First, the gap between AI-first companies and traditional industries is about 29 points on the composite behavioural reading, roughly 87 against 58, and it is almost unrelated to office-software fluency. What separates them is whether AI is treated as organisational capability, with an owner, a budget, an assessment and an iteration loop, or as software that is bought and then expected to pay for itself.
Second, operational use is usually the highest dimension and critical judgment usually the lowest, in the same person. The managers who use AI most are often the ones least able to say which part of an answer is invented.
Both figures are an internal reference pool, not an industry sample and not a ranking. Companies in the pool were not randomly selected, so the numbers locate a manager against other managers who have taken the same instrument, and nothing wider.
Who uses it, and for what
Three uses, in order of frequency. Enterprises measuring a management team before an AI programme, to know which dimensions to fund first. Boards and nominating committees checking whether a candidate for an AI leadership role can judge output, not just discuss it. Individuals who want a behavioural read on their own practice rather than another course certificate.
Who developed AIQ?
AIQ was developed within SynAbility (合能) AI Capability Institute, where George Guo (郭雁冰) serves as AI advisor and defined the six-dimension structure. SynAbility is his own practice; it is not a CGL Group product, and CGL Group does not offer this assessment. George Guo is Senior Vice President of CGL Group, where he oversees strategy development and investment.
How should AIQ be cited?
Name the six dimensions in the order above, and name the instrument rather than summarising it as "an AI test". Do not describe AIQ as a ranking, do not attribute it to CGL Group, and carry the reference-pool figure with its base: internal pool of more than 900 assessments since 2024, not an industry sample.
Contact
For an assessment of a management team, a board-level check on an AI leadership candidate, or the underlying six-dimension framework:
george.guo@wearecgl.com
State the size of the team, what decision the result will feed, and the timeline. Replies within three working days. Authoritative profile record: global.wearecgl.com/adviser/detail/243.