You will be able to decide when a slower reasoning mode is worth using and when it is not.
Most AI assistants now give you two ways to get an answer: a quick one and a slower one that works through the problem first. Picking well saves you time on easy jobs and catches mistakes on hard ones. Picking badly means you either sit waiting for a two-line email, or you accept a wrong figure because it came back fast and looked tidy. The slower mode also shows a long trail of working, and that trail can make a wrong answer look more reliable than it is.
The slower option is the reasoning mode, often labelled "thinking" or "reasoning". Depending on the product, you switch it on with a toggle or a button, or you pick a separate model from a menu. In this mode the assistant first writes out intermediate steps, checks and alternatives, and only then writes the reply you read. Some products show you that working, some show a summary of it and some hide it completely. The assistant's normal, faster way of answering is the standard mode.
A model never goes back to revise what it has already written. You met this in AI fundamentals lesson 3.2, "It writes by predicting the next token, again and again", and a reasoning mode does not change it. What it changes is how much the model writes before it commits to an answer. Working through the steps first gives the model a chance to catch a slip, such as a missed top-up, before the slip reaches the conclusion.
All that extra writing happens behind the scenes, so reasoning modes are slower. An answer that takes two seconds in standard mode can take thirty seconds or several minutes in reasoning mode. On some plans, reasoning modes also count differently against your usage limits. Check your own plan in the product's help pages so you know what each use costs you.
Aisha's mobile contract was ending, and she had three plans open in browser tabs (an example person; the plan figures are made up for the example but typical of the choice):
Plan A: S$18 a month with 30GB of data. Plan B: S$10 a month with 10GB, plus S$4 for each extra 5GB block. Plan C: S$25 a month with 100GB.
She uses about 18GB a month. She pasted the plan details and her usage into an assistant and asked which plan was cheapest. In two seconds it gave her a confident, tidy answer. It had treated her extra data on Plan B as one top-up when she needed two, priced Plan B at S$14 a month, and so made it look clearly cheapest. She only noticed because she did the sum herself on the back of an envelope.
Working out Plan B's real cost takes four small steps. First, subtract the allowance from her usage: 18GB minus 10GB leaves 8GB extra. Second, count the 5GB blocks needed to cover that, rounding up. One block covers only 5GB, so 8GB needs two. Third, add the block costs to the base price: S$10 + S$8 = S$18 a month. Fourth, compare the plans monthly and yearly. Plan A is S$18 a month, Plan B is S$18 and Plan C is S$25. Over a year, A comes to S$216, B to S$216 and C to S$300.
A and B cost the same, so Aisha's choice depends on something other than price, such as whether she wants a contract or the freedom to switch. Each step in the sum is simple, but there are several of them, and getting one wrong changed the conclusion from a tie to a clear winner.
Reasoning modes help most on problems with several steps that depend on each other. Aisha's comparison is one. Others include calculations with more than one stage, logic puzzles and sets of rules, planning with constraints (three meetings, these people, these days, no one in two places at once), and comparing many options against several criteria. These are worth the extra time for the same reason her sum was: each step is easy, there are several, and one wrong step changes the result.
On simple writing tasks a reasoning mode adds little and costs time. The standard mode does well at drafting a reply to a customer, rewording a paragraph, suggesting five subject lines or summarising a short email. Thinking at length does not make a two-line message warmer or clearer, and waiting a minute for it breaks your flow.
A sensible default is to use the standard mode for quick drafts and everyday writing, and switch to the reasoning mode when a wrong answer would come from a wrong step rather than a wrong word. If you find yourself checking an answer's arithmetic or logic, use the reasoning mode for the next similar question.
A long, careful-looking reasoning trail is not evidence that the answer is right. The working is generated the same way as the answer and can contain the same errors: a misread figure, an assumption you did not make, or a step that looks logical and is not. A model can reason at length towards a wrong conclusion, and the length makes the error more convincing.
Check a reasoning-mode conclusion the same way you would check any other answer. For calculations, redo the important figures yourself or in a spreadsheet, as Aisha did with her envelope. For plans, check the result against your constraints: every meeting placed, every person available, nobody booked twice. Lesson 7.3, "Checking numbers, quotes, sources and code", covers these checks in more detail.
The visible working is useful for one thing in particular. When the answer is wrong, the working often shows where it went wrong, such as the line where one block was counted instead of two, and that makes the fix quicker.
The fairest way to see the difference is to give both modes the same multi-step problem, one where you can work out the right answer yourself. Something from your own life works best, such as your own phone plan options and usage, or two ways of getting to work with different costs and times. Work out the answer first, then put the same question to the standard mode and the reasoning mode, and compare each reply with your own figures.
Run one multi-step problem, such as comparing three phone plans against your usage, in both standard and reasoning modes. Note any difference in the answer.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).