I have a soft spot for Zero Hedge, you've gotta love folks who've predicted 1,837 of the last 2 recessions. There's a three part ... zoological research expedition on their contributors over at FT: 123.
Do you need a randomized medieval city? You bet I do!
My previous attempts to put a submarine datacenter equidistant between Europe and the US were not, I thought, given adequate attention but here's how you build a datacenter on a boat.
I don't really know what to make of this article given my loathing of the trolly problem but What if Buddhist monks trained AIs? I'd like to read the actual thesis and my opinion of Harvard has taken a bit of a dip since they misspelled 'off' but it seems interesting.
In more 'automating biology' news, a CT scan is a whole lot of data that is traditionally reviewed by humans to answer specific diagnostic questions. What if you feed it to a neural net instead? Answer: p < 0.00005 5-year mortality predictions. The authors dropped into a reddit thread on the paper and seem pretty confident that there's much more granular predictions to be had. Neural nets like a lot of data and CT scans are generated at a pretty good rate already so I'm excited to see what happens next.
Neural networks can also learn to negotiate with humans. Has anyone patented a chat bot that 'negotiates' your health insurance rates the day after your CT scan? No reason.
Did you notice the little blurb in there about how the bots ended up diverging from english into 'their own language'? No? Well, they did and balls have zero to me to me.
At least it's not as bad as lead, which is amazingly bad for society. Good thing we phased that out in the...what's that? Oh. Never mind (Only applies to piston-engines which is ~2% of aviation. Still!).