AI making us think less? What happens when we outsource our thinking?

Published on 31 August 2026

From writing emails to decision making, AI can take over tasks that once required us to think. A neurologist explains how this may affect memory, attention and critical thinking.

At a glance

• AI has moved beyond helping us find information — it can now write, summarise, analyse, synthesise and make recommendations.
• While there is currently no evidence that AI directly causes brain atrophy, concerns remain that repeated use of AI to perform mental tasks may reduce the cognitive effort needed to keep our brains actively engaged.
• The impact may differ across life stages. Young people are still developing core thinking skills, while older adults benefit from continuing to challenge the brain and build cognitive reserve.

Integrated into everyday realities, Generative AI has moved beyond helping people find information. It can now write, summarise, analyse information, connect ideas and make recommendations — often within seconds.

That convenience has changed how everyday tasks are approached. Instead of working through a problem independently, users can increasingly ask AI to do it for them.

Dr Jennifer Hung, Consultant in the Division of Neurology at National University Hospital (NUH), cares for patients with a wide range of cognitive conditions, from young-onset memory complaints to severe dementia.

She has observed that by constantly letting chatbots solve daily problems, individuals may bypass some of the mental effort the brain requires to maintain cognitive function.

The problem may not be AI itself, but what we stop doing

Technology has long helped people reduce the amount of information they need to remember or process. GPS means there is less need to memorise routes. Search engines make information readily accessible without requiring people to remember where they first encountered it.

Generative AI goes even further. It can take a task from the initial question through to the final answer — finding information, interpreting it, connecting different ideas and producing a conclusion.

This encourages cognitive offloading, where mental tasks are delegated to an external tool. While mostly helpful, the concern arises when AI routinely takes over tasks that are important opportunities for learning or practising cognitive skills.

Dr Hung noted that AI differs from tools such as calculators and GPS because it can perform aspects of higher-order thinking, rather than automating a mechanical task.

Outsourcing this synthesis removes the mental practice needed to build genuine human expertise.

What happens when we outsource understanding?

Nearly 65 per cent of Singapore’s working-age population uses generative AI tools, ranking the nation second globally. As these tools become part of everyday work and learning, another concern is emerging: the difference between producing an answer and understanding it.

Dr Hung shared that AI chatbots can create an “illusion of competence” by producing polished and convincing work with relatively little effort from the user. As a result, people may believe they understand a topic more deeply than they actually do.

There are, however, consequences to the overreliance. “Because the brain skips the effortful process of learning and problem-solving, critical thinking and error-checking skills may weaken over time, increasing the risk of accepting inaccurate or biased information without questioning it,” she shared.

Memory retention is affected

Memory involves several processes, including working memory, encoding and retrieval. Encoding is particularly important because information needs to be processed and stored before it can later be recalled.

“When you expect an AI to always hold or retrieve information for you, your brain skips the encoding phase entirely,” Dr Hung explained. This phenomenon is known as digital amnesia.

For instance, consistently relying on AI to retrieve meeting summaries removes the brain’s need to actively process and record those details in the first place.

Children and older adults may have more to lose

The potential impact of relying heavily on AI may also differ depending on a person’s stage of life. For children and teenagers, the concern centres on cognitive development. Their brains are still developing skills (e.g. reasoning, memory retention, attention and problem-solving).

In Singapore, over half of eight year olds already use AI tools like ChatGPT. That figure surges to over 90 per cent by age 13.

AI can be a useful learning tool when used appropriately. However, if it routinely replaces the effort involved in solving a problem, forming an argument or making sense of a text, children may have fewer opportunities to practise the cognitive skills needed for independent learning.

Optimising long-term brain health for seniors

For older adults, the considerations are somewhat different. Maintaining cognitive reserve, the brain’s resilience to age-related changes, becomes increasingly important with age.

Retirement, for example, can reduce some of the everyday activities that previously required planning, decision-making and social interaction.

“We do see a greater risk of cognitive decline in patients who are newly retired and have remained physically and mentally sedentary,” she noted.

Technology should not replace activities that challenge the brain. Learning new skills, staying socially connected, exercising regularly and pursuing mentally demanding hobbies can all contribute to healthy ageing.

Not less AI, but more active thinking

AI is likely to become an increasingly familiar part of how people work, learn and communicate. The goal is therefore not to avoid using it, but to use it in a way that preserves opportunities for active thinking.

Dr Hung recommended the “Brain-First” principle: to complete the most cognitively demanding parts of a task independently before using AI to refine or challenge the result.

For example, before asking AI to write a report, a user could first identify the main argument, organise the key points and decide what evidence is needed. AI can then be used to review the structure, identify gaps or suggest alternative perspectives.

The same principle can be applied to problem-solving. Users can first work through the problem and then use AI to challenge their reasoning.

In this way, AI becomes a tool for extending thought rather than replacing it.

In consultation with Dr Jennifer Hung, Consultant, Division of Neurology, Department of Medicine, NUH.