AI & Tech Roundup: Stephen S. Roach on who’s winning the US-China AI race, Mariana Mazzucato and Fausto Gernone on why AI should help fund creative labor, and more
Also with Gary Marcus, Christopher Marquis, Peter G. Kirchschläger, and Sebastian Vogelsang
AI’s Reliability Crisis
Gary Marcus
NEW YORK – OpenAI’s celebrated o3 model recently said that I died, and that was not even the worst part. Although o3, the product of years of research, is one of the latest, greatest large language models (LLMs) available, the obituary that it wrote for me (at the behest of a podcaster who was running an experiment to test one of my AI predictions) was filled with hallucinations. It offered a lovely, but fake, quote from AI pioneer Yoshua Bengio – “We should be grateful for critics like Gary … they keep us honest about what’s still missing” – and it claimed that I once publicly debated Geoffrey Hinton, the “godfather of AI” (he has always declined my offers).
Of course, o3’s obituary was mostly correct. It did mention my PhD, some of my books, my AI company (which was acquired by Uber), my testimony on AI to the US Senate, and so forth. But if we know anything for sure about generative AI models like o3, Grok, and ChatGPT, it is that they are often amazing but rarely reliable.
To take another example, my friend Harry Shearer recently sent me an AI-generated bio that miscategorized him as British (he is American, born and raised in Los Angeles) and misidentified his character in the film This Is Spinal Tap (he played Derek Smalls, not David Stanhill).
Anyone who has used these tools knows how common such errors are. When I asked ChatGPT to draw a map of the United States and highlight those states with ports and above-average per capita incomes, it generated several different, inconsistent answers, none of which was correct.
This commentary is from the newest issue of our magazine, PS Quarterly: Post Americana. Subscribe to PS Premium to read the rest of the issue and unlock our magazine archive.
AI Should Help Fund Creative Labor
Mariana Mazzucato and Fausto Gernone show how today’s innovation economy exploits the very people it relies on and propose a fairer system.
Who’s Winning the US-China AI Race?
Stephen S. Roach thinks the outcome will likely come down to which government invests more in basic theoretical research.
AI’s Corporate Shell Game Must End
Christopher Marquis calls for state-level reforms to public-benefit corporations, given the lack of federal oversight in the US.
AI Is Not Your Friend
Peter G. Kirchschläger warns against trusting algorithmic surrogates that cannot grasp the ethical implications of their actions.
Europe Can Build Its Own Social Media
Sebastian Vogelsang thinks policymakers are overlooking the potential for developing platforms on decentralized, open protocols.