There was some excitement online recently around “Freysa”, a contest which combined large language models (LLMs) with some sort of blockchain technology. The basic idea was for participants to interact with an LLM to try to get it to do a thing that it had been instructed not to do. The trick was that the cost of each attempt increased after every submission (up to some maximum), as did the potential reward.
Somewhere on social media I encountered the recent story that the Tony Blair Institute for Global Change, (a large, avowedly centrist think tank created by Tony Blair), put out a report estimating the potential impact of AI on jobs in the public service, but that ironically, they arrived at their estimates by basically just asking GPT-4. While this does have a certain kind of poetry to it, it’s not exactly what we think of as a reputable methodology.1
A whole genre of podcasts seems to have emerged recently, taking the shape of interviews with CEOs of AI companies, or similar. Although I have not listened to many of these, they mostly seem to be pretty abysmal, both because of the amount of hype they involve, and due to the lack of meaningful specifics.
That being said, of the ones I’ve heard, the recent interview with Dario Amodei of Anthropic on the Ezra Klein Show seems to me to be vastly better than the average. In part, this is because Klein is a smart, curious, and well informed interviewer, who asks a lot of the right questions, and pushes for details when responses get too mushy. In this case, it also helps that Amodei seems much more straightforwardly honest about limitations than most spokespeople in similar positions.