We Can Do Better

OpenAI's recent announcement on ChatGPT's bluer content is bad news

As a young man, I had an infection in my brain. I got better, mostly, but I thought that was a brutal experience. Starting an online platform might be on par, give or take some hyperbole.

I'll admit that I'm not well versed in web culture. Just starting this newsletter led me into almost every conceivable pitfall, including impersonation, hacking attempts, and spammers piggybacking on my reputation. But I've worked for a startup in a massive business ecosystem — healthcare — so I figured I'd have some carry-over skill in launching a platform.

Physicality should have been the obvious problem. I was a door stepper, back in the day. When I needed something to happen, I looked for whoever could help, and wrangled my way into a room with them. This was remarkably easy because healthcare is a bureaucratic mess, and such corporate disasters love meetings. All you have to do is volunteer yourself into them and turn up with coffee. (Once, just out of curiosity, I decided to see how many "rolling" monthly meetings I could stay silent in without being dropped off the list. Infinity, it turned out).

With such easy access to people, you can achieve tons during the few minutes around a meeting. Catch someone on the way in, and the knowledge that they’ll be stuck with you for the next hour will make them more sociable than normal. Catch them on the way out, and politeness might force them to agree to meet one-on-one later. Face time is human psychology's default interaction mode.

It's also the only one I'm good at. I'm so bad on the phone that sales teams used to look at me like I'd shot an albatross. And despite writing half decent articles (I hope), my emails read like a robot co-authoring with a chimpanzee. Online outreach was always going to be a challenge, particularly since anyone important is effectively forced to close themselves off.

If you're a big player, any publicly available communication channel to you is inundated. Even I get more inbound messages than I can deal with, so they go unread. I consider silence to be kinder than sending AI-generated responses, but I may be a minority. LinkedIn, for instance, is a bucket of bots these days. Send any more than a few dozen messages — which is tiny, if you're launching something — and you'll realize that lots of them reply quickly and vacuously. I just don't see the point.

I get that big players receive lots of communication, and I realize that much of this is from annoying automated outreach tools. But let's cut the bull crap: if you're using a bot to reply to messages you're receiving from other bots, who's benefiting? Answer — whoever's selling the bloody things. What a racket? I wish I'd thought of it.

And is this really the state of AI? Have we witnessed the birth of an era-starting technology — the first meaningful information retrieval tool over our vast unstructured data — only to have it reply to social media messages and post comments?

Sadly, I can answer my own question with a resounding “no!” OpenAI recently announced that ChatGPT will allow smut for paying customers. I have no idea what the marketing for this will be, but I imagine it'll be something like, "if you can imagine it, we can help you blast a hole in your laptop". (Actually, that's quite good. I'm claiming it as intellectual property).

Messaging, social media, and sticky tissues. Fifty-six years ago, a man walked on the Moon. I'll leave that hanging.