If 2021 was the breakthrough year for mRNA vaccines, then 2022 may be the breakthrough year for artificial intelligence. So far there have been major advances in text generation and image generation, and now investor Nat Friedman is predicting big developments in AI personal voice assistants. More innovation will surely follow.
With so much in the works, it is worth reviewing some basic principles of how AI is likely to affect the US and global economies.
First, more AI does not necessarily mean fewer jobs. Automation and mechanization have been around for centuries, and they create jobs as well as take them away. The job created may be direct, as in robotics production and the surrounding infrastructure, or indirect, as when AI-supported animation is sold and marketed.
Prior to the pandemic, both Japan and the US had full employment and decent living standards. Both were also two of the most mechanized societies in the world and had plenty of robots.
Nonetheless, AI may bring big changes in compensation. Software can already write simple news articles, especially for standardized events such as earnings reports or sports scores. The wages of journalists will fall accordingly, while those of specialists who work on AI will rise.
But the benefits of AI do not accrue only to those in the technology sector. AI makes many goods and services cheaper, and that in turn benefits the poor and disadvantaged. If software routes packages and shipments more efficiently, then transportation costs will be lower. If software and AI programs help economize on the use of electricity, then it will be easier to mitigate climate change. As computational biology improves health care, the sick will benefit.
The people who least need AI are the super-rich. They already can hire armies of servants to manage their obligations, schedules, and so on. They do not need to economize on the use of human labor. The rest of us do, whether directly or indirectly through the businesses we patronize.
Another benefit for lower-income groups is that current manifestations of AI do not usually displace the jobs of the poor. Many poor individuals hold jobs in the service sector or perform manual labor. Those tasks are either hard to automate (a robot gardener?) or, because wages are low, less profitable to automate.
It may be true that the costs of AI in the labor force — displaced jobs — are more visible than the benefits of AI — new jobs and lower prices. So it’s not surprising if AI is not entirely popular.
AI also is going to require a lot of job retraining. More jobs will require new skills that involve working with software, sometimes in the form of AI. It’s not like everyone will have to know the intricacies of neural networks, but often this retraining will be more daunting than learning how to use the new espresso machine at Starbucks.
On average, better educated individuals are more skilled at retraining. So increasing the importance of retraining will have some significant inegalitarian effects, namely that the lesser educated may fall behind or be re-employed at lower wages. This is the real worry about AI, which is not the same as “the robots taking all our jobs.”
Ideally, continuing progress in AI will create a lot more jobs in teaching, training and retraining. That is the optimistic scenario. Still, given that so many Americans do not finish four-year college — even when doing so could lead to better jobs and higher wages — retraining probably won’t be so easy, either. Penalizing those who are bad at retraining may not be exactly the tonic American society needs right now.
Thinking about AI in terms of investments, one obvious play is to try to select the companies that will be making the breakthroughs. More generally, highly successful AI, other things being equal, probably means higher prices for land and natural resources. There will be a lot more economic activity, which will consume more energy and take up more space.
The US and the world economies are going through the early phases of a fundamental transformation, one that will be in large part driven by AI. When the history of our time is written, these will be seen as the years when everything got started.
Bloomberg
