The Possible Impact of Artificial Intelligence in the Labor Industry
The short answer to how more productive machines will impact employment and wages, under the classical model we learned in class, is none. Machines, part of capital, are complementary with labor in the production function-an increase in capital increases the productivity of labor-shifting the marginal productivity of labor, labor demand, up. With fixed labor, this increased demand will increase wages. Employment will stay the same; it is fixed. However, this model is very unrealistic. There is unemployment; labor can increase.
A more realistic model will include real wages in the labor supply function. The higher real wages, the more people would want to work a standard upward-sloping supply curve. A shift upwards in MPL would increase employment and real wages, but real wages would not increase by as much as if labor is fixed-the vertical shift in the latter case is greater than the diagonal shift in the former. While under both models, wages would increase, and employment too under the latter, the rise of AI can significantly alter those models.
The Cobb-Douglas function of complementary inputs would be outdated, as while AI increases the productivity of very high-level workers in competitive fields (e.g. lawyers the lawyers I’ve talked to and worked with are not afraid at all of Al because of a logical fallacy that one party is always going to have the better case, and if software help both parties, the party with the better case is always going to win), it actually causes demand for labor in other sectors to shift down, as it is a direct substitute for labor. If the substitute is cheaper, people would substitute to that product. AI is likely to be cheaper than wages in the long run, and even likely in the medium run.
What happens to workers who are replaced by AI? Job retraining into sectors that cannot be replaced by AI would be prohibitively expensive, and the number of sectors shrink as AI progresses. Even doctors and surgeons are under immense risk, as are the very programmers that are so fond of programming AI.
Government spending on job retraining would decrease public savings, which could have the added effect of stunted investment in AI, but private investment is huge in AI, Competition in AI with America’s peer adversaries i.e. China on military and public security applications, in which America is years behind, unlike in civilian applications, would in fact spur government investment on AI. Universal income would be nice, but it would totally destroy the work ethic, and the self-improvement of humanity. There are two solutions: integrating AI with humans i.e. cyborgs, or stopping research in all civilian applications for Al by law.
Returning to a bygone era would be welcome to many, but the decorum and norms of society have changed so much that most would rather see robots rather than bell boys, camera measurements rather than tailors, self-driving cars rather than valets, and Siri rather than secretaries. Therefore, the former option is the most likely, and best, solution. However, the income gap in implementing silicon with the organic is something that will take much more than one page to discuss.