Worried about and AI "paperclip maximizer" that decides humans are in the way of its goal? What if we already have that? What if corporations and their need to maximize profit are "slow AIs"? A must-read from @timoreilly https://t.co/0UnOFOWfPL (my hubby)
Reminded by several examples this week that pre-2022 web scrapes will be the last of the non-synthetic human-generated content for LLMs, meaning that pre-2022 training data will be valuable like non-radioactive steel.
Jack Smith: "It is very important for me to note that the defendant in this case must be presumed innocent until proven guilty beyond a reasonable doubt in a court of law. To that end, my office will seek a speedy trial in this matter."
George Takei Recalls His Childhood in a WWII Internment Camp. "I could see right outside my schoolhouse window the barbed wire fence and the sentry tower as I recited the words 'with liberty and justice for all.'" https://t.co/CBen8Q7Grr
@justinhendrix I am beginning to think this a ploy for more AI funding. First create a fictitious catastrophic problem and then solicit funding to solve the non existent problem and save humanity.
While it is certainly tiring
to worry
about the virus
and intolerance
and the climate
and artificial intelligence
it is even more tiring
to realize that you need
to worry
about all of those things
at the same time.
@katypearce In the pocket of the blazer / suit coat you only wear to conference presentations that you finally sent to the dry cleaners after several years (and is still there since you forgot about it).
Thus, it follows, I argue that if you want to create a technology that doesn't immediately (or, very quickly) get exploited by white supremacists, then you have to DESIGN for that. It should be part of the process.