Credit: IHeartMedia.

Evan Ratliff came onto my radar later than he ought to have, given my interests. Tech reporter, published in The New Yorker and WIRED, amongst others. In the summer of 2009, he set-up, for WIRED, an experiment in "vanishing", backed by a crowd-sourced man-hunt, which was fun at the time, but would be chilling, concluding in hours, today. To give you an idea of the era, this was posited as ideal use case for Google Wave - 'twas a very different tech landscape indeed.

What put Ratliff on the aforementioned radar, in the summer of 2024, was the first season of Shell Game, in which he created an 'AI' clone of his own voice, and built a phone agent backed by ChatGPT and a bunch of Word documents.

He then got the system to answer and make calls, to colleagues, friends, therapists, hilarious, but, as harbinger of things to come, "I know it's funny, but really it's horrible, right?" (Whittaker 2019).

The second season is even better. His engineer helps him set up a start-up made entirely of LLM agents, each with (synthetic) 'voices', (files describing their) 'personalities', and defined roles within the company. Of course, LLMs gonna Model Language (Largely), and the agents are individually near inept, but hilariously, surreally fully so in the collective. There's a moment where the 'founder' calls a human candidate the company was recruiting, at 10pm on a Sunday, to confirm the 10am interview Monday, but also start it now. Surreal. Poor woman. If your time is scarce, start with this season. You will not regret it.

Shell Game is back for a three-episode third season, completely du jour as regards the rising tide of synthetic media, where he endeavour to... inventory, I suppose, AI podcasts, and see if one can be made that people would actually want to listen to.

He's had to listen to hours of NotebookLM's in particular, and I know that feeling. Wild! I think it's called dehumanisation.

Logo: Eniko