I’m getting really annoyed with the doomsday predictions about AI.

I recently watched a few clips from Steven Bartlett’s conversation with David Friedberg about AI and jobs. The thing that got me wasn’t that I agreed with every point. It was how quickly a prediction can start sounding like a fact just because it’s repeated with enough confidence.

I’m happy to be proved wrong. AI is moving quickly, and some people will lose work because of it. But “AI will take all the jobs” is not analysis. It’s a prediction, and a very convenient one for the people selling the future.

Before we frighten ourselves with another forecast, I think we need to get a few things straight.

01. AI can sound human. That doesn’t make it human.

AI can sound kind, curious, embarrassed, even worried. But producing the language of an emotion is not the same as feeling it. Research into a current AI model found emotion-related patterns that influenced its behaviour. The researchers were clear that this did not show the model had subjective feelings.

The same care is needed with “reasoning”. AI can solve problems that look like reasoning. That is impressive. But it can also give a fluent, confident answer that is wrong. The system’s ability to produce an answer doesn’t mean it understands the world as we do, or has experienced what that answer means for the people involved.

AI can process images and audio, and generate faces, voices and expressions. But that is different from having a body, a life, or the experience of being in the room. It has never noticed that the conversation everyone is having isn’t the one they need to have.

That doesn’t make AI useless. It makes it different. We’re better off understanding that difference than pretending we’ve hired a new colleague who happens to live inside a computer.

02. AI will probably take the work we already complain about.

My prediction is that AI will first take a lot of the tedious, repetitive and wasteful work people do because nobody has fixed the process.

Copying information between systems. Formatting reports. Producing the first draft of a routine response. Searching, sorting and summarising. Chasing updates. Entering the same information in three places.

Most of us can name tasks we’d happily never do again. AI can help with those because it works quickly across large amounts of information, doesn’t get bored by repetition, and can prepare a rough first version for someone to check.

But there’s a question we shouldn’t skip: what does the organisation do with the time it saves?

It might improve the service, give people room to think, or let them focus on more useful work. It might also cut the team and ask the people left behind to do more. Calling that “productivity” doesn’t make it a good outcome for the people doing the work.

The International Labour Organization estimates that one in four workers worldwide are in occupations with some exposure to generative AI. It concludes that most jobs are more likely to be transformed than made redundant. That’s useful evidence against claims that whole occupations will simply vanish. It is not a promise that nobody will lose their job.

“AI replaces tasks, not jobs” is a useful distinction, but it can become a comforting phrase. Jobs are made of tasks. Remove enough of them and you may need fewer people.

There’s an awkward thought here: some work looks valuable because it takes time. But being busy isn’t the same as doing the work where your judgement and experience matter most.

03. The opportunity is to let people do more of the work that needs people.

I’m less interested in whether AI will help us produce more. We already have plenty of ways to make more things. What matters is what people can do with the time and capability it creates.

Imagine a patient with a disease so rare that their doctor may never have seen another case. A system could help search across medical records, research and genetic information to find patients with similar symptoms or variants in other countries. Researchers are testing AI tools that can identify “patients-like-me” across clinical datasets. That match might help a medical team see a pattern, consider a diagnosis or investigate a treatment option. It is not the AI diagnosing the patient or prescribing a therapy. But it could help turn “we don’t know what’s happening” into a more informed conversation about what to investigate next.

That’s the kind of possibility I find genuinely exciting: not a machine replacing the doctor, but helping people find connections that are hard to see when the evidence is scattered across the world.

In creative work, AI can generate options. Experience helps us decide which one fits the brief, the audience and the moment. In leadership, AI can organise information. A person still has to make the decision, explain it and take responsibility for what follows.

The machine might prepare the answer. A person still has to decide whether it is the right answer, what it leaves out and who will live with the consequences.

If it becomes easier to produce work, judgement becomes more important. The question shifts from “Can we make something?” to “Is this any good? Does it mean anything? Is it right for us?”

There is a catch. AI will not automatically give people more time. Companies may simply expect fewer people to produce more. The opportunity has to be chosen and designed. Otherwise, we’ll automate the drudgery and keep the pressure.

04. History is useful. The comforting version of history isn’t.

Desktop publishing changed the printing industry. Computers eliminated some typesetting and composition jobs, while shifting other work into offices and creating demand for new production skills. That wasn’t painless, and it didn’t protect everyone whose work had changed.

So I don’t think history proves that AI won’t take jobs. That would be a silly conclusion. It does remind us that technology changes work in ways that are rarely as simple as “everything disappears” or “everything gets better”.

Who benefits? Who loses? What new work appears, and who can access it? Those questions matter more than the familiar story about carriage drivers or the claim that every previous technology worked out fine. The transition can create new possibilities and still leave people behind.

I’m not asking anyone to stop taking AI seriously. I’m asking us to take it seriously enough to stop confusing dramatic predictions with evidence.

AI will change work. Some people will be hurt by that. But the outcome isn’t written in the technology alone. It will also be shaped by the choices businesses make and the value we place on human judgement.

The doomsday predictions make the future sound like something happening to us, as if nobody has any say.

We do have a say. We can choose which work to automate and what to do with the time it creates. The machine can help us get through more work – do we want that?. That still leaves the harder question: what work is worth doing, and what do we want people to be there for?

For brands, the more AI can produce, the more important it becomes to have a clear point of view that people recognise and trust.

This article was written with the assistance of AI. The ideas, thinking, topics and structure are mine, as are parts of the writing and the way I’ve expressed them.