It is not being clever, and it is not being sceptical

Two misreadings do most of the damage. The first confuses critical thinking with intelligence — the assumption that sharp people reason well naturally. But generating arguments and testing them are different skills, and the second is where expensive mistakes live: a clever person can build a persuasive case for a decision they have already made.

The second confuses it with scepticism, which quickly becomes a personality rather than a method. Critical thinking is not a mood about other people's reasoning; it is a procedure you apply to your own, and the discomfort is that its first target is you.

The failures it is actually there to prevent

Strip the jargon away and critical thinking exists to stop a small number of recurring errors:

• Deciding from a single vivid case. One bad experience with a supplier becomes a policy. The story is memorable; the base rate is not.
• Treating confidence as evidence. The loudest person in the room is not the best-calibrated one, and the two get confused in every meeting.
• Judging a choice by how it turned out. Good decisions produce bad outcomes. Reward only results and you teach people to be lucky.

Each of these is a habit, not an intelligence problem, which is precisely why training can shift them.

Why experience alone doesn't build it

Experience teaches you the situations you have already met. What it builds is pattern recognition — exactly the faculty that fails when a situation is genuinely new: a new market, an unfamiliar supplier, a technology nobody in the room has deployed. Critical thinking is what you fall back on when the pattern does not match, which is why it sits unused until it becomes a deliberate practice.

The cost of a wrong call has gone up

Decisions that used to be local are now expensive. Procurement cycles are longer and more visible, data protection obligations have teeth, and a poorly reasoned approval paper can sit in a queue for weeks before anyone discovers the assumption underneath it was never tested. The faster an organisation moves, the more it pays for the reasoning it skipped.

AI made the answer cheap, which moved the value to the question

This is the shift that matters most. A plausible, well-formatted answer is now free and instantaneous. What is scarce is the ability to tell a good answer from a confident one — to check the claim, find the assumption, notice what the summary left out, and ask the question nobody asked.

Which means the practical value of critical thinking has not gone down with AI. It has moved. If your team's job is increasingly to review and judge machine-assisted work rather than to produce a first draft, then judgment is now the core skill, not the supporting one. Training that only teaches people to generate output is training for the part that just got automated.

There is a data discipline attached to this, too. Anything confidential you put into an AI tool is a PDPA decision, not a productivity decision — keep client material, personal data and unannounced plans out of tools you do not control.

Separate the claim from the evidence for it

Most disagreements are muddy because the claim and the evidence arrive in the same breath. "We should switch suppliers" is a claim. "Our current supplier was late twice last quarter" is evidence. They get tested separately, and evidence can be true while the claim follows from it not at all.

The habit is mechanical: write the claim in one line, then list what is offered in support of it. The gap between those lists is where the real discussion is.

Ask what would change your mind

This is the most useful question in the set, and the hardest to answer honestly. If nothing could change your mind, you are not reasoning toward a conclusion; you are defending one. Ask it early, and ask it of yourself first — a recommendation with no stated falsifying condition is not a recommendation, it is a preference with supporting documents.

Find the base rate before the story

Attention is drawn to the specific and the dramatic, and reasoning built on a vivid example is reasoning built on the least representative data available. The correction is to ask, before the story is told, what usually happens in situations like this: how often do projects of this shape overrun, and what proportion of these applications are approved?

Base rates are boring, which is why they get skipped, and reliable, which is why they belong first.

Judge the decision separately from the outcome

Keep two ledgers: was the reasoning sound given what was known, and how did it turn out? They diverge in both directions — a reckless call can succeed, a careful one can fail for reasons nobody foresaw.

Organisations that track only outcomes teach their people to be lucky and to hide the rest. Judge the process at the time of the decision, and record it before the result arrives — which is what the next section turns into a method.

Write the decision as a question

Not "we need a new system" but "which of these options should we commit to for three years, given that our volume may double?" A decision written as a question can be answered and revisited. One written as a statement can only be agreed with or resisted.

Choose the criteria before you look at the options

This step most changes outcomes, and it is the one people skip. Evaluate options first and you end up with criteria reverse-engineered to justify the one you already liked — a process that feels like analysis and functions like decoration. Fix your criteria and their weights first, in writing, then bring in the options. It is uncomfortable because it works.

Score it — and then interrogate the score

A simple weighted matrix is enough for most workplace decisions:

Cost over three years | 3 | 4 | 3 | 2 |

Implementation effort | 2 | 2 | 4 | 3 |

Fit with current team | 3 | 5 | 3 | 2 |

Reversibility | 2 | 2 | 4 | 1 |

Risk of vendor lock-in | 3 | 2 | 3 | 4 |

The table's value is not the total in the last column. It is that you can see which criterion is doing the work — and whether your weight for it is honest. Change one weight from 3 to 5 and the ranking often flips, which tells you the decision is really a debate about priorities, not options.

Record what you expect, before you know

Before you commit, write one sentence: what you expect to be true in six months if this was the right call. It takes thirty seconds, and without it hindsight rewrites the reasoning. With it, you have a dated record of what you believed — and a team that can disagree about results without disagreeing about history.

Practice on live problems

You can read a book about weighted criteria and still weight them dishonestly the moment the decision is yours. The gap between understanding a method and running it under pressure is where training earns its fee — which is why the useful question about any course is not what it covers but how much of the day is spent practising.

Feedback that sees your blind spot

Everyone has one recurring reasoning failure. It is almost impossible to find your own, because it is the shape of your thinking rather than a gap in your knowledge. A skilled facilitator watching you work a live decision can name it in an afternoon, in a way no amount of reading will.

WSQ and funding: what it changes

Many courses in this space are WSQ-accredited, which matters twice over: employers can claim available funding, and SkillsFuture Credit may offset fees for eligible Singaporeans and permanent residents. If you are sponsoring a group, ask how the claim is filed and what each participant must complete — attendance and assessment requirements are usually conditions of the funding, not optional extras.

Check the current fee and funding position on the provider's own course page — funding tiers and eligibility change, and a course fully subsidised last year may not be this year.

Ask for the decision frameworks, not the theory list

A course that advertises six theories of reasoning and no framework you can use on Monday has been built to fill a syllabus, not change a practice. Ask the provider to name two or three frameworks participants take away and how they apply to everyday decisions — a procurement call, a hiring decision, a project review. Ask as well how many participants there will be and how much of the day is spent practising with feedback: a hall of eighty with a workbook is a briefing, not training.

Bring a real decision to the course

The highest-return preparation is to arrive with a live decision you are actually facing — the messy version, not a cleaned-up one. Participants who work on their own problem leave with a decision and a method; those who work on the trainer's case study leave with notes.

Treating it as a one-off event. A single workshop produces enthusiasm and, within a month, nothing else — unless a practice survives it. Attach it to something already happening: approval papers, project reviews, the Monday decision meeting.

Measuring satisfaction instead of behaviour. "Everyone enjoyed it" tells you the facilitator was personable. Measure whether the papers got sharper, whether options are listed before criteria, whether anyone asks what would change their mind.

Sending only the junior staff. A junior employee who tests a senior's assumption without cover learns quickly not to. The method holds only if people with authority use it on themselves, visibly.

Critical thinking at work is a small set of habits: separate the claim from its evidence, ask what would change your mind, find the base rate before the story, and judge the decision separately from the outcome. Written down, that becomes a method — question first, criteria before options, interrogate the score, record what you expect.

A course is worth the money when it gives you practice on live problems with feedback from someone who can see your blind spot, and when the practice survives the workshop. It is not worth it when it is a theory briefing, attended once, by the people with the least authority to use it.