The divide that rejecting AI could create

There is a kind of resistance to artificial intelligence that turns into a judgement about the person using it. Simply admit that you have worked with AI and the suspicions begin: perhaps you cannot do it yourself, perhaps you have not made an effort, perhaps what you have produced is worth less.

The conversation starts with a technology and ends with putting people into categories.

I believe that sorting people this way can have consequences far greater than an uncomfortable conversation. The world seems to be dividing into those who embrace AI and those who reject it. And stigma can push that divide to the point where it becomes a difference in skills, opportunities and income.

The social penalty is real

Four experiments published in PNAS in 2025, involving 4,439 participants, found a social penalty for using AI: people described as AI users received lower ratings for competence and motivation.[1]

What changed was how those people were judged in the experimental scenarios. Knowing which tool they used was enough to damage their image.

A professional may set out to find ways to do their job better, only to discover that they also have to weigh up what others will think if they explain how they did it.

Microsoft and LinkedIn found a related signal in their 2024 survey: 52% of those using AI at work were reluctant to admit using it for their most important tasks.[2]

The incentive is easy to understand. If explaining yourself damages your reputation, staying quiet begins to look like a good strategy. You can learn to use AI and, at the same time, learn to hide it.

Looking human at any cost

Suspicion about authenticity brings that logic into writing.

A text can arouse distrust by being too neatly structured. A clear sentence can seem too polished. The absence of errors can become evidence against its author.

If a writer goes so far as to introduce imperfections to avoid being mistaken for a machine, that logic reaches the point of absurdity. Deliberately spoiling a text to make it seem human turns mistakes into a badge of authenticity.

But that badge can also be manufactured. A machine can be asked to produce a clumsy sentence. An excellent text may have been written by a person without any help.

Suspicion ends up taking the place of reading. Instead of examining an idea, people look for clues that will let them categorise the person presenting it.

And when the label determines the value of the work in advance, explaining the method starts to carry a social risk.

The pressure to pick a side

For someone starting out in law, that atmosphere can be particularly harmful.

Think of someone who wants to learn to work with AI. They ask questions, experiment and compare results. If the people around them associate those practices with laziness or poor judgement, the message is clear: if you want to be taken seriously, some things are best kept out of sight.

They may carry on learning in secret. They may also turn away to avoid being judged. In both cases, social pressure has shaped a decision that should be guided by curiosity and professional judgement.

Someone starting out needs to explain what they did, receive specific criticism and understand how to improve. Stigma turns that conversation into a test of belonging: you are one of those people who use AI.

That is how sides form. Using a tool ends up looking like a statement of identity. Rejecting it can become a way to win approval within the group.

The problem arises when keeping that approval means giving up on learning.

The advantage grows while we argue

My sense is that we may be witnessing the birth of a socioeconomic divide in real time.

Those who bring AI into their work begin moving along a learning curve. They try out tasks, spot errors, refine their processes and discover what they can achieve. Each attempt provides knowledge for the next.

Those who reject that practice outright stay off that path. Time passes for both, but the experience they gain is different.

That is the advantage the early adopters have. I believe that this time they could gain a very substantial competitive advantage in the short term. They can learn to handle more work, develop services and reorganise what they do while others are still deciding whether they find using the tool acceptable.

That difference can translate into the ability to take on projects, fulfil assignments and offer solutions that were previously beyond their reach. It can also mean higher incomes and greater freedom to choose how they work.

For a young legal professional, time spent learning can become a skill they bring to their professional life. If pressure from the group draws them away from that practice, the cost goes beyond an uncomfortable discussion. It can affect the opportunities they will be able to recognise and seize.

That is exactly where stigma comes in: it can make someone abandon learning to retain the approval of others.

A divide that can remain hidden

Nor does the divide necessarily match what people say in public.

Those who use AI in secret can also build their capabilities. Their silence preserves the appearance that everything is the same, even as the way they work changes.

That makes the divide harder to see. A group can appear hostile to AI while including people who are already moving ahead with it. Those who take the group's rejection at face value may miss out on learning that others are already acquiring.

That is why it concerns me that using AI could become a source of shame. It can punish openness, reward concealment and discourage precisely those who still have time to get started.

This has practical consequences. The more some people learn, the more ground those who decide to join them later will have to cover.

A professional should be able to say they are learning to work with AI without first having to defend their intellectual worth.

Social suspicion may seem like a passing reaction. The gap in capabilities it leaves behind may last much longer.