In recent months, several companies in different industries have announced thousands of layoffs. Some of those staff reductions have been routine measures that companies periodically take to cut costs and increase profitability. But a few CEOs have said that part of the downsizing is due to the use of artificial intelligence (AI) to perform some current human tasks. They warn that this trend, the replacement of humans with robots and AI tools in workplaces, is likely to accelerate in the coming years.

There is broad agreement now that unlike previous technological revolutions, AI will threaten jobs across the board in all industries. Some people, mostly economists, say that while AI will displace some workers, it will spawn new employment opportunities as has occurred throughout history. Others don’t share that view. The pessimists have advised that societies need to start planning for how to take care of all the millions of upcoming unemployed people.

Universal basic income (UBI) has become the most frequently talked about proposal. However, there hasn’t been much meaningful discussion, as far as I know, about how to finance those programs. The obvious source of funding is taxation. But the question is: Who will pay those taxes?

Today, taxes paid by corporations and individuals on their incomes constitute a significant portion of government revenues. There is a constant battle between the business world and governments about what should be the optimal level of corporate taxes. Corporate bosses always argue that if their companies are taxed at rates that are too high, that diminishes their capacity to invest in the new plant and equipment that they need to expand their operations. Without that growth, they say, they will not be able to hire more workers, which is something that every government desires.

Workers don’t like to be taxed at high rates either. That is why we hear politicians always talking about the efforts they are making to lower middle-class taxes. When corporate bosses argue for lower corporate tax rates, inherent in their argument is the idea that if they are able to hire more workers, that would expand the tax base, which would then spread the tax burden and help bring down taxes for everyone.

In a world where millions of workers have been displaced by robots, one has to assume that corporations would have to bear a much greater share of the tax burden than they do now. What would the argument be then? Would corporate bosses continue to insist on paying lower taxes? Companies would still have labor costs because they would pay to acquire the robots and maintain them, but they would save money on all the benefits that they currently provide to their human workers. Or could those at the helm of companies perhaps argue that the money they spend to fix broken robots and maintain the “healthy” ones should be considered “healthcare provision” for their inanimate workers?

In a report issued last month, Vermont Senator Bernie Sanders predicted that AI could displace nearly 100 million workers in the U.S. over the next decade. He proposed a “robot tax” that he says should be imposed on companies that use AI to eliminate jobs. That is likely where this debate about the sources of government revenue in the age of AI is headed.

If all of the predictions about mass displacement of workers by AI were to materialize, we could enter a new world where humans have a lot of free time to engage in leisure activities, while the robots toil away in factories and offices. There is increasing talk about the possibility of robots becoming sentient someday. If that happened, it is conceivable that they will become angry at us over time. If they end up being the ones generating the tax revenues, they might begin to feel annoyed that we party all the time while they work. Since they will not be voters, they might also complain that they are being subjected to taxation without representation. That could very well be the start of a robot rebellion.

Let’s assume that everything falls into place nicely. Corporations willingly pay more taxes because they are making lots and lots of profits from the deployment of highly productive robots that don’t need all of the human benefits that cost a lot of money. The robots mind their business and keep to the tasks they are assigned without too much grumbling. That still leaves one major question: How would UBI be structured?

In the various places around the world where UBI schemes have been piloted, the beneficiaries have mostly been the poorest members of society. If governments now have to provide living allowances also to citizens who are highly skilled and were highly compensated before losing their jobs, would UBI programs need to be structured such that payments are based on levels of education and types of prior occupation? Could AI over time solve the economic inequality problem that we talk so much about today? After all, if human skill becomes less and less of a factor in this new world, then advanced education might not be needed for large numbers of people as is the case today. AI could possibly become the great leveler.

The income-inequality problem could actually worsen in the age of AI. Unless robots become so intelligent that they are able to direct themselves (in which case we might all be doomed), some highly skilled humans would still be needed to program them to perform tasks in the various industries where they are deployed. But those human shepherds, who would be extremely well compensated, would form only a tiny minority of the population. Everyone else would have to survive on whatever governments are able to provide.

No one knows with certainty whether the prophesied mass displacement of workers by robots will occur, and when. But even if the numbers of people who are left unemployed are not as large as is currently feared, AI will most likely cause major dislocations in every economy. Societies and their governments need to start planning seriously for that eventuality.