One Man's notebook this week.
Nice to see that @marsedit@mastodon.social has escaped squircle jails…
One of the deep problems with algorithmic social media is the incentive to figure out how to game the system: “growth hacks” as a certain constituency likes to call them.
Some of them are surprisingly akin to religious rituals. You have faith that they work, but not a lot of evidence…
The challenging thing about running a training course on Ai over four weeks, is that things can change dramatically between sessions…
Still, kudos to Google for digitally watermarking their AI-generated images.
Video generation is getting good scarily fast.
There are some weird, weird sheep in there, though.
Turns out, AIs are a humourless bunch.
(I’m mainly talking about satire and cartoons here. Maybe the role of humans in the AI age is… taking the piss.)
This is a sobering look at OpenAI’s funding. No sniff of a profit for at least a decade, according to HSBC forecasts.
A lot of non-industrial AI use is predicated on it being a cheap and easy solution to certain problems. If customers weren’t being so heavily subsidised by investors, when does that equation start breaking down?