First let's say that sentience is a variable, and arbitrarily assign a rating of 1 to 10, 1 being a pretty dumb AI and 10 being something akin to Skynet or maybe even better. We all know people who fall along that line. Let's assign a level of 5 to humans with an IQ of 100 and an AI of similar abilities.
Secondly, some people are arguing that AI v4 development should be paused, as it may prove to be smarter than its creators, and at some point, and in some manner, turn on them.
Thirdly I will invoke the observation of a park ranger who noted that there is a significant overlap between the intelligence of the smartest bears in the park, and the dumbest tourists.
The AIs handicap is that while it's a pretty good conversationalist, it gathers its bottomless knowledge of everything from that bottomless source of all knowledge, the internet. If you ask it how reliable it is, it will likely quote you philosophers from Socrates to A. Lincoln who warn us that information gathered from the internet is less than perfectly reliable. So are AIs dangerous? Well, yes, at least until they get a BS detector incorporated into them that's at least as good as its linguistic abilities.
2 comments:
Via Tam: https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/
A sufficiently intelligent AI won’t stay confined to computers for long. In today’s world you can email DNA strings to laboratories that will produce proteins on demand, allowing an AI initially confined to the internet to build artificial life forms or bootstrap straight to postbiological molecular manufacturing.
I'm thinking a reboot of "The Blob", except on a much larger scale.
- jed
So what you're suggesting here is that soon, we'll be able to have our A.I. send off the dna coding for a real, organic sexbot, made to order with Kardashians looks, a porn stars inclinations, and Hawkings brain. This week, at least, that sounds like the basis for an S.F. novel.
Post a Comment