AI Won’t Replace Everyone. But Who Will Own the Future?

Artificial intelligence is often discussed in extremes. Depending on who is talking, AI is either going to replace almost every human worker or create an unprecedented era of productivity and prosperity. The reality is probably much more complicated. Some jobs will disappear, others will change, and many will simply become more productive. The question is not whether AI will have an impact, but how large that impact will be and who will capture its benefits.

Predicting the future of work is particularly difficult because the effects of technology depend on the industry, the type of work and the economic environment. Instead of making absolute predictions, we should look at measurable things such as employment, wages, productivity, social mobility and the size of the middle class. Those numbers can tell us much more than another prediction about what AI will supposedly do five or ten years from now.

My biggest concern isn't necessarily that AI will eliminate most jobs. It is that AI could concentrate even more economic power in the companies that own the technology. If AI allows a company to accomplish the work of hundreds of employees with a much smaller workforce, the resulting productivity gains have to go somewhere. They could become higher salaries, lower prices and shorter working hours—or they could simply become higher profits for companies and shareholders.

There are already examples of companies laying off large numbers of employees because they believe AI can perform much of their work. But there are also cases where those expectations haven't materialized and companies have ended up hiring people again. AI can be extremely useful without being capable of replacing an entire human role. There is a huge difference between saying that AI can automate 40% of someone's work and saying that the company no longer needs that person.

I've actually seen both sides of this personally. At one of a good friend's jobs, more than 100 people were laid off as part of an AI-focused strategy. The company had a relatively focused and profitable product and was able to build a powerful AI framework around it. At another job, the product is considerably more complex, and the company understands that current AI systems cannot simply absorb all that complexity. In that environment, the expectation is actually that more people will be needed.

That is why the comparison with ATMs is interesting. When automated teller machines appeared, many people assumed they would eliminate bank tellers. Instead, ATMs automated certain tasks while humans continued to perform others, particularly those involving interaction, judgment and more complex services. Technology clearly eliminates jobs, but automating tasks is not necessarily the same as automating an entire occupation. AI may follow a similar pattern.

The part that worries me more is what happens if AI becomes dramatically more capable. Imagine that a company can produce twice as much with half as many employees. That could be fantastic if workers benefit through higher wages, shorter hours or better opportunities. But if most of that additional productivity flows to the owners of the technology, society could become considerably more productive while simultaneously becoming more unequal.

This also makes AI regulation particularly important. The companies investing billions in AI have an obvious interest in influencing the rules surrounding the technology. Regulation can protect society from legitimate risks, but it can also unintentionally make it harder for smaller companies to compete if compliance becomes too expensive. In the worst case, the rules designed to control AI could end up protecting the companies that already dominate the industry.

There is another factor worth considering: the enormous amount of investment flowing into AI. The technology is real and its potential is enormous, but expectations can grow faster than actual capabilities. Companies have a strong incentive to demonstrate increasingly spectacular AI achievements because those demonstrations attract capital and increase valuations. That doesn't mean AI is a scam. It simply means that a genuinely transformative technology can also produce an investment bubble.

So, will AI replace jobs? Absolutely. Some companies will discover that they can operate with fewer employees, and some professions will be transformed or partially automated. What I remain skeptical about is the much larger claim that AI will eventually replace essentially all human work. Maybe it will. Maybe it won't. We simply don't know yet, and anyone claiming absolute certainty probably knows less about the future than they suggest.

Perhaps the more important question isn't "Will AI take my job?" but "Who will benefit when AI makes human work dramatically more productive?" Will workers earn more? Will products become cheaper? Will working hours decrease? Will small businesses gain capabilities that were previously available only to corporations? Or will most of the economic gains accumulate among the companies and investors that own the technology?

Those questions can actually be measured. We can look at wages, employment, productivity, wealth distribution and social mobility. Because ultimately, the success of AI shouldn't be measured only by how intelligent our machines become. It should also be measured by what happens to the people who have to live and work alongside them.

There is also a less dramatic, and perhaps more interesting, possibility: AI may not destroy jobs on a massive scale, but instead change who gets to compete. Someone with access to good AI tools can now accomplish things that would have required an entire team just a few years ago. For a small business, independent developer or creator, that can be extraordinarily empowering. But if those same tools are available to large corporations, they can also use them to increase their competitive advantage, acquire smaller competitors and reduce costs on a much larger scale.

That is why I don't think this discussion should become a battle between "good AI" and "bad AI." Technology doesn't have an economic intention of its own. What matters is how we use it and the incentives surrounding it. A tool that allows one person to do the work of three can be liberating if it means fewer working hours and better income. But it can be harmful if it simply means that three people lose their jobs while a company captures almost all of the resulting benefit.

And perhaps that is the real uncertainty. We don't know how capable AI will eventually become, but we also don't know how companies, workers and governments will respond as those capabilities increase. It is entirely possible that ten years from now some of today's predictions will look ridiculous, both the optimistic ones and the apocalyptic ones. That's why I would rather look at actual results than buy into either narrative. AI will probably change many things. The bigger question is whether that change will make the economy work better for more people, or simply make those who already have the most resources even more powerful.