AI Could Be the Best Thing That Ever Happened to Humanity

The public conversation around artificial intelligence has become strangely narrow.

Every day we read another story about layoffs, restructuring, hiring freezes, automation, productivity gains, cost reduction, and companies “reallocating resources” towards AI. The language is polite, but the message is not difficult to understand. Businesses have found a technology that may allow them to reduce wages, benefits, workspace, administrative overhead, management layers, and operating friction.

So when people hear that AI is “transforming work”, many translate it into something simpler:

AI is coming for my job.

That fear is not irrational. It is already happening in some sectors. Companies are not waiting for some perfect future model of AI adoption, they are acting now. They are cutting, freezing, reorganising, redeploying, and investing heavily in automation. The job losses are immediate and specific. The promised new jobs are often described in broad, abstract, future-tense language.

That is the uncomfortable gap in the optimistic AI story.

We are told that AI will increase productivity. We are told productivity will increase growth. We are told growth will create new jobs. We are told history has seen this before. But the serious question is rarely answered clearly enough:

What exactly are these new human activities?

Not in theory. Not in a consultancy diagram. Not in a 2030 forecast. What are people actually supposed to do when their current work is automated before the new work has appeared? This is the missing middle of the AI employment debate.

Forecasts may well be right that AI creates as well as destroys jobs. The World Economic Forum expects millions of new roles by 2030, even after very significant displacement. Goldman Sachs has argued that AI could ultimately support productivity and job creation, while also warning that hundreds of millions of jobs globally are exposed to automation and that displacement during the adoption period is a real risk. McKinsey has described a future in which agents and robots can perform large parts of today’s work activity, while humans remain necessary for judgement, supervision, exception handling, creativity, and redesign.

So the debate should not be reduced to the crude question: “Will AI destroy jobs or create jobs?”

The better question is:

Will AI be used to reduce human value, or to release it? That distinction matters.

A company can use AI to automate the present. It can take existing workflows, existing reports, existing customer service processes, existing administrative structures, existing internal approvals, existing sales funnels, and make them faster and cheaper. That may improve margins. It may please investors. It may reduce headcount. It may make the organisation look modern. But it may also make the company intellectually smaller.

Because improving an existing process is not the same as asking whether that process should still exist.

This is where the AI-agent debate becomes more interesting. AI agents are often presented as autonomous workers: systems that can complete tasks, coordinate steps, produce outputs, monitor information, make recommendations, and perhaps even act across software tools. That is useful. It may be extremely useful. There is no virtue in forcing humans to spend their lives moving information from one system to another, writing routine summaries, checking predictable records, producing unnecessary reports, or managing administrative noise.

The best argument for AI agents is not that they replace humans. The best argument is that they could free humans from the kind of work that stops humans from thinking. That is the opportunity, but it is not automatic.

If AI agents are used only to cut cost, the result may be fewer people doing more supervision of more machines, under more pressure, with less time to reflect. That is not liberation. That is acceleration and acceleration alone is not progress. A business can become faster and still become outdated.

It can automate customer service without asking why customers need so much service in the first place. It can automate report production without asking whether the reports influence any meaningful decision. It can automate internal approvals without asking why the approval structure is so heavy. It can automate recruitment screening without asking whether it is hiring for the future or merely replacing yesterday’s job descriptions. That is the danger of using AI only inside the existing frame.

AI can optimise the process. Humans must still ask whether the process should still exist. AI can improve the machine. Humans must still ask whether the machine is the right one. AI can help us do things faster. Humans must still ask whether those things are worth doing.