Today, Bill Gates published an extensive essay on his Gates Notes website, demonstrating the depth of the Microsoft co-founder’s considerations regarding artificial intelligence’s societal implications.
Primarily aligning with the Responsible AI movement, Gates suggested that Pacing the Frontier, an open letter from AI professionals advocating for a deceleration in AI development, presented a sound concept, though he voiced doubts about its long-term viability.
While enthusiastic about AI’s potential advantages in scientific and medical fields, he articulated concerns regarding its effects on employment, topics commonly explored in Anthropic’s policy discussions.
However, he introduced several truly novel concepts that, if adopted, could significantly reshape ongoing debates. One such initial proposal is a “robot tax”:
Currently, employers incur payroll taxes when they employ individuals. In contrast, acquiring a robot typically allows for an immediate business expense write-off. This taxation structure subtly encourages the substitution of human workers with machinery.
Implementing such a tax would modestly decelerate the shift away from human employment and generate funds for professional retraining programs and an enhanced social safety net.
Additionally, he advocates designating specific occupations as “Human Reserved,” thereby prohibiting AI involvement in particular tasks. This intriguing concept appears straightforward for policymakers to implement:
Designating roles as Human Reserved could be justified by economic considerations. For instance, such a measure might prevent widespread job displacement in situations where machines taking over a specific function would affect many individuals unable to readily transition careers. It’s unrealistic to expect a 55-year-old with a lifetime in construction to find fulfillment in a new role within an elder care facility.
Occasionally, the choice to make a role Human Reserved will stem from non-economic factors. In healthcare, for example, a robot could technically deliver the devastating news of an incurable illness. However, it ethically should not.
The scope of Human Reserved roles will adapt over time — for example, we should contemplate reserving certain jobs now and gradually integrating AI over several years or decades, with a pledge to maintain a portion of these positions.
It is also worth noting that both these proposals would significantly impact the profitability of leading AI laboratories, potentially explaining their limited public discussion thus far.
Furthermore, numerous questions remain regarding the precise entities responsible for establishing these regulations and the specific content of such rules, should they be implemented.