3 questions about Gen AI

June 25, 2023

Question 1: How will it affect the labour market? One possibility is that it will increase the returns to "knowledge work" (coding, writing, etc) and therefore make those jobs more productive and highly paid. The other possibility is that it will reduce the need for humans doing those services and therefore depress those wages. What combination of the two will happen, remains to be seen.

(Prediction: Because written / audio / visual content will be easier to generate and technical knowledge more accessible, personal characteristics will matter more. There's already bias favouring charismatic, attractive individuals and affluent well connected people. I predict this will get worse.)

Question 2: How will the provision of foundation models evolve? They may become a utility provided by a 10-20 companies - like mobile phone operators, grocery supermarkets, or retail banks. In this case others will build applications and specialisations on top of them. Alternatively models may become more like services by the hundreds (like restaurants, schools, moving companies) where each company actually builds, maintains, and updates a separate model. Currently the balance is for the former - but it's too early to tell.

Question 3: what new professions and markets will be created? Other than selling models/API-calls for others to build applications (question above), what markets will be created for new revenue? It's easier to see the opportunities for cost-savings: deploying Gen-AI to help coders, call center operatives, doctors, etc. But it's harder to see new markets being created (like the smartphone created a market for digital apps, or YouTube for video content creation, etc). What new products will you buy in 2 years? What new professions will be with us by 2024?