Arthur Holland Michel

Ok, what now?

Friendly reminder that if you don't have to pay to use something that costs $3million/day, you may be paying for it by other means.

#chatGPT

Among SBF's Future Fund donations are $300,000 for the author Brian Christian to finish a book manuscript and $190,000 for "a reputable technology publication" to hire...2-5 student interns.

https://ftxfuturefund.org/our-grants/?_area_of_interest=artificial-intelligence

Today in our ongoing series, Mistakes Computers Make that No Human Ever Would:
"KFC Uses Nazi Pogrom to Promote Fried Chicken in Germany"

https://www.vice.com/en/article/m7g4vq/kfc-germany-kristallnacht

Techy question: is there a way to rate the 'predictability' of the timing and character of an AI error? From a policy and safety perspective, my sense is that errors that happen at predictable times and cause predictable results are far less of an issue than wildly unpredictable errors.

Ukraine’s digital transformation minister says 'autonomous killer drones are “a logical and inevitable next step.”'

Startling quote in this great write-up on autonomous weapons from the AP.

https://apnews.com/article/russia-ukraine-war-drone-advances-6591dc69a4bf2081dcdd265e1c986203
.
.

Still, the stadiums were built by slaves.

What's the over-under on the tech slump being good news or bad news for AI ethics?

Is Mastodon supposed to be a vehicle for comedy? Just want to calibrate expectations from the outset.

Hi! I've just written a story for The Atlantic about how AI can (and in many cases can't!) sort and moderate the endless video feed that's become the dominant mode of social media content these days:

https://www.theatlantic.com/technology/archive/2022/11/tiktok-instagram-video-feeds-ai-algorithm/672002/