## Issue 67:Woodpeckers Working on a Big Tree

4 min readMar 4, 2024

Hello all,

A newsletter post today! Kind of. I’ll offer a few quick ideas, because today these thoughts are pecking at me like woodpeckers working on a big tree.

Subscribe now

TOP OF MIND

I’m closing in on the final draft of my second novel, the sequel to Surrender. It’s called Resist. It has another editorial round ahead before it is released into the wild. I’m aiming for a preview soft launch in a couple of months, with the real launch after that. This has taken a lot of bandwidth, as you might imagine. The third book in the series, titled Liberation, is in the early stages.

I also have a non-fiction project, with a codename for now: Hypercities. I’ll be looking for an academic press for that. Anybody on this list who went to Yale, Princeton, or MIT, I’ll be getting in touch to see if you went to any reunions lately or know the editors at those university presses.

Stunningly, USC, the folks I teach for, who have been supportive, and who recently promoted me to assistant professor, they don’t have a university press. I was surprised to learn that, but at least my students call me “Professor,” which is nice.

FAKE OUT LOL

What’s fake, and what’s real on the internet?

News this week about AI-generated food appearing on DoorDash. This is troubling if you like hoagies, heroes, grinders, or chicken parm. 404 Media broke the story:

Dozens of Ghost kitchens, restaurants that serve food exclusively by delivery on apps like DoorDash and Grubhub, are selling food that they promote to customers with AI-generated images. It’s common for advertisements to stage or edit pictures of food to make it look more enticing, but in these cases the ghost kitchens are showing people pictures of food that literally doesn’t exist, and looks nothing like the actual items they’re selling, sometimes because the faulty AI is producing physically impossible food items.

There’s a new book out about crypto by Chris Dixon, a VC. It became popular fast. Apparently, like many crypto markets themselves, the folks behind the book gamed the system to make the book rise in the NYT bestseller lists. Their method: Just buy up a shitload of copies.

The world of internet fakes is getting stronger and weirder, but humans are fighting back in small ways. A web designer and activist is offering graphics that you can put on your website to identify it as made by a human. Expect to see more of this watermarking behavior, the more sophisticated among it being software-driven.

HOW I’M USING AI

From the alarmist tone I adopt here sometimes, you may think I am completely against using AI. Not the case. AI is an excellent research assistant, with the right guardrails.

I use (and pay for!) a web browser called Kagi. When I ask a question in the search bar, Kagi responds with a summary answer in plain language, sourced from reliable links. I use Kagi because Google results have become garbage. (Reference below.)

I use (and also pay for!) an AI tool called Podium.page which does a good job transcribing and summarizing podcast recordings. It tries to write promo copy which, sadly, comes out at the level of the intern’s intern. But it’s a start and every little bit helps.

HOW I’M NOT USING AI

From the above, you can see that I’m using AI that bases its results on material that I put into it (like podcast recordings that I made), or from links openly available on the web. I am not using AI to make videos, drawings, or marketing copy that derives (was stolen) from artists. I have tried Claude.ai and OpenAI’s ChatGPT and become dismayed at the different ways they steal from humans. Automattic, the parent company of WordPress and Tumbler, just cut a deal with OpenAI and Midjourney, to gobble up our blogs and images for their AIs to learn from. Sad. But I think you can stop them! I’m working on getting my other sites, hosted on other providers, protected from getting crawled by bots. In a later post, I’ll let you know if I’m winning the battle against these bots that want my work.

CORRECTION

In the last newsletter, I wrote about an apparent George Carlin deepfake. Some creator types claimed they had prompted an AI to write an entire George Carlin comedy monologue in Carlin’s voice. Then the truth came out. They wrote it themselves and got an AI to imitate Carlin’s speaking voice. There was no AI involved in the creation of the monologue, just human idiots.

DISCLAIMER

This newsletter was written in haste, like a pack of hungry dogs was chasing me as I ran away cradling a wireless keyboard, typing as fast as I could. Typographical errors, poor metaphors, broken grammar, and strained examples are not the responsibility of the author. I will find someone, or some software, to blame as soon as I have time.