Will AI Growth Reach the Developing World?
Here is the abstract from a new NBER paper by
This paper explores how economic growth driven by AI will impact the developing world. Our analysis is based on a growth model of a developed and developing economy where AI services are traded on the world market, and may substitute for or complement different labor inputs in production. Preferences allow for the possibility that demand for services provided by AI has an upper bound. The model predicts that developing countries may not benefit as much from AI-driven growth. First, they may lack complementary skills needed to fully utilize AI domestically. Second, they may be priced out of world markets due to their relatively lower productivity. We use our model to simulate how growth and welfare are affected by making workers more complementary to AI, subsidizing compute, speeding technology diffusion, and providing cash transfers. Numerical simulations of the model suggest that the potential for developing countries to be left behind – or having some workers lose in absolute terms – is substantial enough that researchers should study the impacts of AI on developing countries in real time.
I am more optimistic than the authors. For one thing, in their model AI improves the productivity of legal services in say Africa, but not food. That mix seems unlikely to me. AI already makes rainfall prediction and crop management more productive, and it might over time help create more “miracle crops.” A few other points:
1. I do not expect wealthy and poor countries to converge, and the gap might even increase with AI. But the poorer countries can sell more of their resources to the wealthy countries, and are likely much better off anyway, compared to no AI. If you read the authors carefully, they do not (seem to?) deny that, yet it is framed as being “left behind.”
2. You might think that AI and robots discourage the export-oriented industrialization for today’s poor countries. In my view, that was not going to happen anyway, admittedly that is a subjective judgment.
3. The poorer economies typically have an excess of politics and corruption, even if government spending as a share of gdp is relatively low. This suggests “AI lawyers” will not put “African lawyers” out of business soon, even if the AI lawyers are very good. Practicing law is bundled with politics to a very high degree. That fact is unfortunate, but it implies a high degree of job protection and a centaur model, rather than full work displacement.
4. Will the poor countries satiate in AI-provided services, such as legal services, as is the case in some versions of this model? That seems unlikely, as more growth can boost the demand for legal services more or less indefinitely. And so so many people in those countries right now cannot afford decent lawyers. That prospect of satiation seems very far off indeed.
5. Rather than having a single world market for AI, why not have tiered markets, where the poor countries buy lower-quality AI services but at lower prices and also they avoid being shut out of the prime market?
6. The autors (p.11) hold TFP fixed for many parts of the economy. Of course that will lower the gains from AI.
7. p.44 argues that developing economies are becoming priced out of AI inputs already. I would be shocked if that prediction turns out to be true, rather I expect smartphone usage (as opposed to mere cell phones) to continue to increase at a rapid pace.
So overall there is way too much negative framing in this piece, and the negative framings seem deliberately chosen, rather than falling out of a more basic or primitive model.