I just got asked for my talking points for this meeting at Alpbach on Artificial intelligence and the future of work. Here they are with links to the evidence or full argument.
My key points are:
- AI is not really a significant discontinuity, but rather an acceleration of a process that's been going on for thousands of years.
- AI will lead to more rapid change in the workplace, but whether that leads to joblessness depends more on inequality / wealth distribution than AI.
- Therefore employment depends on financial governance / redistribution, which can be tricky because AI/ICT facilitate transnational wealth extraction, including trillions of micro barters for information that are not denominated and therefore are hard to tax.
- Here is a great example of automation not causing joblessness: there are fewer tellers per bank branch due to AI/ATMs, but there are more tellers altogether because there are now more bank branches since they are more profitable. But their jobs are more people oriented, less counting oriented. (Example due to Bessen (2015), but the linked Autor paper on employment rocks.)
- The idea that robots will take "all the jobs" is insane: there are far more things we could be doing for each other, better services we could all be receiving. Sustainability and inequality are the two big problems; both are solvable but require substantial political and societal will and understanding.
- My key recommendation wrt AI in the workplace: it is important that a company is organised around a core skeleton of directly connected human workers, so that insights and opportunities can be passed between empathic humans, even from the customer to the chief executive if necessary, though in all likelihood empowered individuals will be able to solve many problems. But disintermediation through technology leads to brittleness and makes it impossible to capture both problems and opportunities that may arise that are outside the expected business process.
- On another EU-specific political note, the legal system depends on a human being culpable and punished, so there should be no "legal persons" that are entirely synthetic -- it is not appropriate to reward a company for fully automating part of their business process by capping their tax and legal liability by allowing machines to go bankrupt.
|I see the world remarkably similarly to Tim O'Reilly's (right) new book, WTF? picture credit: Dan Bayer|