We Are Social and Meltwater released Digital 2027: Singapore today, 5 November 2026. The report says 89.9 per cent of Singapore internet users aged 16 and above used an AI tool in the past month, ahead of the global figure of 85.2 per cent. Only 34.6 per cent say they feel comfortable using such tools.

e27 calls that a 55-point gap between habit and confidence. The report adds some texture. Nearly half of Singapore internet users used ChatGPT in the past month. Finding information is the top reason for use, at 63.4 per cent. Only 8.1 per cent pay for an AI subscription. And 73.4 per cent of online news consumers worry about what is real and what is fake online.

The gap is real, and it is not a tooling problem

The easy reading of these numbers is that people need more training on the tools. Better prompts, a course on ChatGPT, a workshop on which model to pick. I don’t think that is the problem.

Look at what the numbers actually say. Usage is high. Comfort is low. People have the tool in their hands and use it every month, mostly to look things up, and they still don’t trust what comes back. Nearly three in four worry about what is real. The tool is already in use. What is missing is the way of working around it.

I have seen this in my own companies. The technology was the easy part. Changing how people think and work was the hard part, even for a tech team. Give someone an AI tool with no frame and they will use it the way the survey describes: ask it a question, get an answer, feel uneasy, and either paste it in anyway or quietly stop using it for anything that matters.

Mindset, skillset, toolset, in that order

When I teach AI in business I use one framework, in order of what matters most: mindset, skillset, toolset. The survey describes a population that has done it backwards. Toolset first, because the tool was free and everyone was talking about it. Skillset is patchy. Mindset, the part that produces comfort, mostly hasn’t happened.

The mindset is a working frame with three parts. You direct. The AI produces. You check and own the output. Once a person holds that frame, the discomfort has somewhere to go. The AI isn’t trusted to be right, and it doesn’t need to be. The person is responsible for every output that leaves their hands, same as before, and the AI is a fast producer under their direction. “Is the AI right?” is an uncomfortable question. “Did I check it?” is a normal one.

The skillset follows from the frame: how to brief, how to check, how to fix the brief when the output is wrong. The toolset comes last, and it changes. Capabilities outlast interfaces. A course on where the buttons are in this month’s app is out of date by the next release.

What this looks like in a clinic or a marketing team

Take a clinic with a front desk team of three. They use AI every week to draft replies to patient enquiries. Nobody said they could, and nobody said they couldn’t. One of them trusts it too much and sends drafts untouched. Another doesn’t trust it at all and retypes everything. That is the 90 and the 35 in one room.

The fix takes an hour, and it is a frame rather than a two-day course on the tool. The AI drafts. A named person checks every reply against the clinic’s actual policy on appointments, fees and what the clinic can promise. That person owns the reply. The frame also says what the AI never decides on its own, such as a refund or a medical question. Write that down and the uneasy people have a reason to use it, and the over-trusting people have a reason to check.

Same for a marketing team. If AI is producing content with no posture and no checker, the output is generic and the team knows it. That is where the discomfort comes from. Put a person in charge of the brief and the check, and the volume stops being a worry. AI makes volume free, and the frame is what decides whether the volume is worth anything.

The brand discovery numbers point the same way

One more figure from the report. Only 12.6 per cent of Singapore internet users discover new brands through AI tools, and 23 per cent use them as a main source when researching brands. Search engines still lead at 51 per cent. Arthur Altounian, Managing Director of We Are Social Singapore, said the brands that grow in 2027 “will be the most believed” rather than the most visible.

That matches what I see in AI search work. The AI doesn’t pick the biggest brand. It picks the one it understands most clearly and can quote most precisely, and that is something a business can work on.

What a business should do

Write the frame down before you buy anything. One page: who directs, what the AI produces, who checks, and what it may never decide on its own. Then pick the skills that frame needs, briefing and checking, and train those. Choose the tool last, and keep it to one tool used deeply rather than a stack used shallowly. Nine in ten of your people are already using AI. The job is to make them part of the comfortable third.

Source: e27: Nine in ten Singaporeans use AI. Only a third are comfortable with it