If I could talk to the animals, just imagine it
Chattin' with a chimp in chimpanzee
Imagine talking to a tiger, chatting with a cheetah
What a neat achievement it would be!” – Bobby Darin – 1967

The trainwreck that is “Failure — the Podcast” has nearly achieved its eponymous singularity: failure! With no recent episodes to point to, we can’t attract guests. Not to mix metaphors, but who’d want to go down with this ship? Think of Paul McCartney’s recent appearance on the now-defunct “Late Show.” Do you think he’ll ever book a gig again? Dr. Andy Clark, our last victim — er, guest — was about to be saddled with that same yoke, but we took pity. The podcast guest-o-sphere had few prospects to offer, save a motley assortment of self-publishing authors looking for airtime. We are desperate, but we still have a smidgen of pride.

Thankfully, Mark had an idea: we could pay Copilot and Claude to be our guests. Sure, they might have done it for free, but then we’d have to wrestle with confidentiality — you know, to protect the integrity this podcast is known for. So we upped the ante and got ourselves some fresh meat. Black gold. Texas tea. (Sorry, couldn’t help but think of the Clampetts and Elly May’s pound cake.)

They spoke well, Copilot and Claude. Each can almost carry its weight in conversation, tempting though it may be to use them as glorified search bots. (Ahem, Mark.) The fact that these two remember context, including prior conversations, is key: it’s no longer Groundhog Day with every new interaction.

In what we hoped was not merely a bout of terminal lucidity, Mark observed that the combination of natural language communication plus context can readily be mistaken for intelligence. That’s not as profound as the Drake Equation, but we think Mark may have something there. This feels like the transition from clanking on the radiator in a cold house to checking in on your smart thermostat from the office.

Speaking of Alan Turing (weren’t we?), he had it all wrong: the true test of artificial intelligence is whether it delivers snark indistinguishable from Jerry Seinfeld’s. By that standard, today’s chatbots are decades away from AGI. What are they good at? Conversation, as already noted. But also, compliments. They dole them out like prospectuses at a Bernie Madoff gathering of A-listers. Between “really great observation,” “interesting question,” and other bon mots, it’s easy to forget that you barely made passing grades in school.

Here are some interesting factoids Mark threw in to meet his word budget for this blurb: more than 20 million people use Claude monthly, a figure that’s up considerably since Anthropic’s kerfuffle with the Department of War, née Defense. That’s real growth, though still well behind Copilot’s 100-plus million active users. What’s especially interesting is how many people regularly engage with two or more chatbots. According to Perplexity, one of Claude’s digital frenemies, two-thirds of Claude users do. It makes us wonder whether these people frenetically poll the chatbots with their every concern: “Is Ozempic safe?” “Do you think I should ask for a raise?” “Why didn’t Cosmo Kramer record a Christmas holiday album?”

What can you, dear listener, get from today's episode? Certainly, we didn't work at conveying anything of value. Still, if you listen even briefly, you’ll realize that AI chatbots are better spoken and more cogent than most of us. That’s a low bar, but it says something — especially in comparison to Microsoft’s AI-adjacent “Clippy” of the late 1990s.

We've come a long way, baby.

Enjoy!