Did We Build a World That Needs AI?

I understand why people worry about artificial intelligence. If your livelihood depends on something a machine might soon do, the promise of greater efficiency can sound a lot like someone explaining why they no longer need you.

But I think there is another part of this conversation that deserves attention.

We talk about AI as though a handful of wealthy technology executives decided what the future should look like and everyone else is being pulled along. Those companies certainly have money, influence and their own interests. Yet that explanation leaves out the people choosing to use these tools.

Some are trying to grow a business. Others need help understanding a document, learning a skill or finishing work before their children get home. Their reasons can be quite ordinary.

I wonder how much of the demand for AI comes from the world we have already built. A world where earning a living, raising a family and keeping up can feel increasingly difficult.

I recently remembered a story about a man who worked at Sears and managed to build a comfortable life selling appliances. It stayed with me because there was nothing extraordinary about his occupation. He had a regular job, and that job supported a comfortable life.

In a 2018 New York Times article by Nelson D. Schwartz and Michael Corkery, the reporters described Stanley Hreneczko, who began working at a Sears store in Michigan in 1965. His pay and the company savings plan helped him buy a home in cash and take family holidays. The article also explained that Sears once devoted 10% of pretax earnings to an employee retirement plan. By the 1950s, employees owned a quarter of the company.

That matters because his security came from more than his ability to sell refrigerators. It also came from the way his employer shared its success.

I am not suggesting everyone had an easy life then. Many people were excluded from those opportunities, and the Sears plan itself had shortcomings. But the story raises a reasonable question: how much security should someone be able to build through ordinary, useful work?

The pressure on earnings is not imaginary. In its 2025 economic survey of Canada, the OECD reported that rents rose 6.3% in 2023 and 7.9% in 2024. Earnings grew 2.2% and 4.6% in those same years. For someone whose rent was rising faster than their pay, working just as hard could still mean falling behind.

Education remains valuable. So do experience, persistence and talent. But none provides a guarantee that the life someone is working toward will be within reach.

Against that background, I can understand why a tool that helps someone do more would find an audience.

For a person with an idea, AI might make it possible to take a first step they could not previously afford. They can work through a business plan, understand unfamiliar code or prepare a proposal. They still need judgment. They still need to check the work. But they have somewhere to begin.

I see that appeal myself. When you are building something, there is always more to learn than there is time available. Getting help with one part of the process can mean finally making progress on the part you understand well.

Research suggests much of the demand is practical. A 2025 NBER working paper, “How People Use ChatGPT,” found that practical guidance, information seeking and writing accounted for nearly 80% of conversations. The study involved OpenAI researchers, a connection worth knowing, but the pattern is revealing. People were using it for help with everyday questions and tasks.

Employers face their own version of this pressure.

Hiring someone takes time and money. Training takes more. When the arrangement does not work, both the business and the employee can pay a price.

I have seen capable people struggle in jobs that did not suit them. Someone can be thoughtful, hardworking and talented while being poorly matched to a particular role. Unclear instructions, inadequate support and unreasonable workloads can make that situation worse.

From the business owner’s perspective, though, the work still needs to be completed. Customers still need answers. Mistakes still need correcting. A small business may have very little room to absorb repeated problems.

That helps explain the attraction of automation. The appeal can be reliability, availability and the ability to handle more work without adding costs the business cannot carry.

Sometimes that will lead to replacing a position. Sometimes it will help the person already doing the job.

A study of 5,179 customer support agents, summarized by NBER in 2023, found that access to an AI assistant increased productivity by about 14%. The gains were substantially larger among less experienced and lower performing agents. In that setting, AI helped people become more effective at their work.

That possibility deserves attention alongside the risk of job losses. A person who struggles to remember every procedure might do well with useful guidance available during a conversation. A new employee might become confident sooner. An experienced employee might have more time for the problems that require their judgment.

There is a familiar economic logic here. The comparison with earlier industrial change is imperfect, but the attraction of producing work faster, more consistently and at lower cost is easy to understand.

What does not follow is that everyone will benefit equally.

A business can become more productive while its employees become less secure. It can save money without lowering prices. It can introduce a tool intended to help workers and then raise their workload until the relief disappears.

This is where the Sears story becomes relevant again. Technology and productivity are only part of the story. Decisions about pay, ownership, working conditions and how success is shared help determine what progress means in someone’s life.

We should also leave room for people to have mixed feelings.

There is nothing unreasonable about using AI to help write a letter while worrying about what it could do to your profession. Someone can appreciate the assistance and still question who controls it, how it is deployed and what happens to people whose work changes.

Using a tool does not mean agreeing to every decision made with it.

I also think we should be more demanding about what counts as an improvement.

I am not convinced that reaching a chatbot immediately is better than spending 45 minutes on hold if the chatbot cannot solve the problem. An instant response has limited value when it sends you around in circles.

But a system that resolves a straightforward issue promptly, and gets you to a person when necessary, could make a frustrating part of life easier.

That is the kind of progress people have reason to want.

My suspicion is that AI has found such a large opening partly because people were already looking for help. Help keeping up, learning, building and managing demands that often exceed their time or resources.

That does not make every application useful or every consequence acceptable. It does mean the demand deserves a more thoughtful explanation than either corporate ambition or public enthusiasm alone.

We helped create the conditions in which these tools became appealing. We also have choices to make about what happens next.

If AI allows us to accomplish more with less effort, I hope some of that gain becomes time, security and opportunity for ordinary people.

I would hate to discover that we had made everyone more capable, only to make a decent life require even more.