Mystery AI Hype Theater 3000 Turns Two Years Old!
Happy Birthday MAIHT3k!
Two years ago today, we went live with our first MAIHT3k live stream! The title was there, but we had no idea it was going to be more than a one-off.
A little back story: We'd both been active deflating AI hype in text-based media for a few years already at that point. In April 2022, in a group chat, we'd been discussing with Meg Mitchell and others what to do about AI hype that comes in the form of bad academic talks posted to YouTube. It was Meg's brilliant idea that the Mystery Science Theater 3000 format might provide good inspiration.
In August 2022, Google VP Blaise Agüera y Arcas posted "Can Machines Learn How to Behave?" on Medium.com--16k words of word salad, armchair theorizing, and navel-gazing conversations with a Google chatbot (i.e. AI hype). Emily initially considered doing a take-down as she had done for Steven Johnson's NYT Magazine piece (the title of which was in our initial graphic for the show) or an earlier blog post by Agüera y Arcas. But his August piece was just so long and Emily, after all, has a day job.
Emily asked in the group chat if anyone was up for giving it the MST3k (actually what she wrote was MST2k, having heard of the show but never actually watched it to that point) and Alex, a fan of MST3k and experienced with streaming, jumped in. We advertised the live stream over social media, lined up a captioner, and got started ...
... only to discover that one hour (actually 40 minutes, by the time we resolved various technical difficulties) was nowhere near enough to wade through all the hype in that one blog post. So we scheduled a second stream and a third, just to clean up all that muck. By then, we were having too much fun AND, of course, the AI hype wasn't stopping. We also figured we could learn a lot by bringing on guests to share their expertise, starting with Johnathan Flowers, Jennifer Lena and Negar Rostamzedeh for our fourth stream taking on the topic of "AI Art".
A few more streams in, we realized that while the recordings of the stream were getting some traction, we'd probably have a lot more reach if we turned the audio into a podcast. With support from DAIR, we were able to hire our amazing producer Christie Taylor, who not only wrangled the early episodes into passable podcasts, but also has helped us level up our audio game, as well as doing all of the million behind-the-scenes tasks it takes to make a successful broadcast. We also commissioned our theme song from Toby Menon and our logo from Naomi Pleasure-Park.
In the past two years, we've been delighted to watch the podcast grow (Emily especially is unreasonably excited to watch metrics like downloads, while Alex wants to resist quantification) and are tickled that the pod has been added to syllabi and suggested as recommended listening for various working groups about AI around the US and beyond. But perhaps the biggest joy has been seeing the podcast become a space where people who have felt all alone in the sea of AI hype, wondering why no one else can see that the Emperor has no clothes, can connect and build community.
Thank you for coming with us on the journey so far. We're also eager for you to read our book, THE AI CON, which we've been diligently editing and will be out next spring. We're looking forward to what the next years will bring!
If you'd like to support the podcast financially, you can contribute to DAIR. We also appreciate support in the form of ratings & reviews on your favorite podcast platform, recommendations to your friends, or boosts on social media.
You can find our complete catalogue of episodes in both video and audio-only on our webpage: dair-institute.org/maiht3k/