John Thickstun

Assistant Professor

John Thickstun, assistant professor of computer science at Cornell University, studies machine learning, has spoken extensively on AI advancement, regulations and infrastructure investment.

Thickstun says:

“It is easy to conflate two sets of two risks, that may have very different likelihoods. On one hand, there are the economic and societal challenges that come with the integration any new technology. On the other hand, there is the risk that ‘this time is different’ and that the challenges posed by AI are unique, e.g., an end to the economic value of human labor, or existential threats to human civilization.

“My position is that this time is not different; the risks posed by artificial intelligence are serious but not unique. Amodei takes some care is his essay to dispute this point, so a careful reader should make up his or her own mind about his counterarguments. It is too easy to conclude that this time is different if we underestimate the societal and economic upheaval caused by other historical technological developments. For example, the role of printing press in 16th century religious wars or the radio in the rise of 20th century fascism, or the labor market pains caused by automation of agriculture and industrialization.

“If we buy into Amodei's frame, that this time is different, then we may focus on the wrong set of risks and we miss the opportunity to learn from past mistakes. Rather than speculatively prepare for fanciful outcomes (end of work, enslavement of humanity, etc.) I think we should focus on mitigating emerging challenges that follow common historical patterns: reorganization of labor markets, demagoguery through new forms of media.”