Writing the Laws Before the Tragedy
On Bill Gates, the “Human Reserved,” and whether rules can arrive ahead of the harm.
I am neither a particular fan nor a detractor of Bill Gates, especially given the more recent reporting about him, but the ideas he shared this week in a lengthy essay on artificial intelligence are worth checking out. His central concern is one I think about regularly, whether AI and robotics will advance faster than our institutions, laws, workplaces, and social structures can adapt.
He argues that the oncoming transition will be fundamentally different from previous technological revolutions like the Industrial Revolution, because AI can duplicate and replace aspects of human thinking. This will impact professions like lawyers, doctors, accountants, programmers, manufacturing workers, and customer-service employees. The difference is that this will not be gradual over years as in the past, but within a relatively short period. Most of the conversation now is about AI as software, but he goes further, to what happens when robots carry that same processing into physical work.
One of his more interesting ideas is what he calls the "Human Reserved." These are occupations or roles that might be deliberately set aside for people, even when technology becomes capable of replacing them. He names childcare and jury service as examples. Others have mentioned some senior care roles being reserved for people. Gates asks whether you want a robot delivering disappointing news about an incurable disease, and his answer is that there is "no technical reason why it couldn't. Yet it shouldn't."
There is a common reassurance in all of this that Gates addresses, that technology always creates new kinds of work to replace what it ends. Some offer the example that stagecoach drivers may have become taxi drivers, who then became Uber drivers. But if there are robotic tools that limit the number of construction workers, does a construction worker late in his career simply become a childcare worker to reach retirement? The new careers also tend to arrive on their own schedule and go to different people.
Gates raises, as a partial solution, the idea of taxing AI and robots. This is because labor may become cheap from the one-time purchase of a robot or system. If the government loses the payroll revenue, there is no money going to a depleting Social Security fund, and there may be a need to support the growing number of displaced workers. In addition to Gates, Sam Altman (OpenAI) and Dario Amodei (Anthropic), both Bernie Sanders and Elon Musk agree that the tax system will need to be reconsidered as AI reshapes the workforce. A universal basic income, once a fringe idea, is now a recurring part of this broader conversation. The Bipartisan Policy Center laid out this week what taxing AI would actually mean, walking through options from taxing AI tokens to taxing data centers to treating an AI as a taxpayer in its own right, while taking no position on whether any of it should happen.
The big question Gates asks is what happens if the technology moves faster than society, opinion, and politics can keep up. The proposals are his, but I have the same questions.
Another question I have is if the men laying the first railways had stopped to suggest speed limits, crossing signals, and rules for who yields at an intersection, would anyone have listened, or would it have sounded like solving a problem that did not yet exist? We tend to write the laws after the tragedy. Gates is suggesting we try beforehand this time.