July 29, 2025 | 06 min read

Similar systems, whether from Google, Anthropic, or X, can perform high quality medical diagnoses, analyze legal problems, and do management consulting, among many other tasks. You do not have to think these outputs are perfect, or fully ready for use, to realize the workplace has changed forever.

But how should a company respond? We are in uncharted territory, but some general principles apply. Most importantly, AI systems can augment the productive powers of individuals who know how to use them.

Tasks that once required a day now take an hour. Someone prompts the Large Language Model, and another person checks that the final product is acceptable. But much of the routine leg work in between is now done by artificial intelligence. In my own work, some of which involves podcasting, I now can prepare for an episode in half the time as before. With good prompts, I can get a digest of background information on the person I am talking to, rather than having to read many books and google around for articles and pdfs.

Put differently: many of us need fewer human assistants.

As a result, many of us can become far more productive. If you know how to work the AI, in essence you have an unlimited number of work aides at your disposal. The quality of that assistance depends upon the task at hand, but AI systems already outperform humans in many areas of knowledge, even if they cannot perform most of the real-world interactive tasks that a job might require. So your assistant is smarter and more knowledgeable than you are, even though that assistant needs direction and evaluation.

Increasingly, we will see scenarios where one person — or a small team — does the work of fifty. Of course, humans still will be needed to oversee editorial checks, do final sign-offs, accept responsibility for the product, show up at meetings, and network with other humans.

Keep in mind that even some of those functions can be managed by applying more AI to the problem. Each AI system is imperfect, but you can ask one AI to check the work of another. Recently, CEO of OpenAI, Sam Altman predicted that we will soon see billion-dollar companies with only a single human employee, who also will be the founder and CEO. That individual will oversee a team of AIs generating the product. For instance, imagine a company that produces marketing copy for other businesses, generated by AI. AI also would do much of the work of reaching out to customers, running the company’s social media feeds, handling customer service, and preparing tax returns. All of this may be two or three years away, but it is no longer science fiction.

My vision is less expansive than Sam’s, as I think such companies will need two or three humans, maybe even five. Still, we will have many more companies and many of those companies will be much smaller in terms of payroll, though perhaps not revenue.

AI is likely to produce both a greater number of firms and far leaner ones. Some of these will be new companies, founded explicitly for the heavy use of AI. But in other instances, companies will evolve into these forms, for instance by allowing attrition of the work force, and gradually turning over more of their routine operations to AI.

More companies will mean more projects and more new ideas. It also means that an entrepreneur will have the chance to try out his or her new idea, without requiring or having to fund large-scale cooperation from others. Our economies will become more innovative and also more experimental. More people will have the opportunity to try to realize their dreams. For these reasons, the advent of powerful AI systems will not lead to mass unemployment. There will be new projects across the economy, and all of them will require human participation. Prompting, directing, and auditing AI will thread through almost every job.

The new jobs of the future might involve running or participating in clinical trials for newly discovered medicines, building out the energy sector to satisfy the growing needs of data centers, entertaining other humans, or being a skilled carpenter or gardener for all the new projects we will be attempting. Building and testing AI systems will be another job growth area, as is already the case. And in many jobs, we may prefer to deal with humans, as when restaurant customers often prefer to have a waiter than to order through an iPad put on the table, even though the latter technology already works well.

Nonetheless, many of these roles will sit inside firms with pared-down payrolls.

If you are a potential founder or worker, you should be learning those skills on your own, no matter what stage in life you may be at. It is time to consider whether you and a handful of partners will become corporate titans commanding fleets of digital assistants.