But I am sure we can make safe AI ... https://www.nytimes.com/2018/01/03/business/computer-flaws.html … https://fb.me/5gC7KlOdg
-
-
Replying to @romanyam
Nobody who deeply thought about it seems to have any confidence that we can make safe AI, but most agree that we cannot avoid making AI, and we should try to make it safe against all odds
4 replies 0 retweets 3 likes -
There is mo such thing in life as "safe". There is only an attempt to create situations where there are as many diverse options that improve life as possible. So, when one fails us, we have other paths available.
1 reply 0 retweets 0 likes -
Replying to @thewiseturtle @romanyam
Turil, safe AI is about existential risk, not diversity of outcome. The attempt to build strong AI is a bit like trying to make a new type of nuclear power plant that has an unknown probability of blowing up the whole planet, but only if it works.
2 replies 0 retweets 1 like -
I see everything we DNA-designed beings do as aiming to generate more options, for more resiliency/fitness. Natural selection weeds out the maladaptive. So adaptive stuff is our aim. AI helps life create more options.
1 reply 0 retweets 0 likes -
Replying to @thewiseturtle @romanyam
Yes, I see that you see it like this. But AI is a different game. AI not just a technology but Life 3.0. It is very hard to create a scenario where it won't replace us.
2 replies 0 retweets 1 like -
But what do you see as the difference between AI and
#MachineLearning? And how close do you think this level of AI is?1 reply 0 retweets 0 likes -
I define intelligence, in any form, as the ability to model three dimensions of perspectives/goals simultaneously while solving a problem. What do I want? What do you want? What does our environment want? Machine learning is one way to take in and process information.
2 replies 0 retweets 0 likes -
Interesting. Current AI has a long way to go then.
1 reply 0 retweets 0 likes
Yes, but nobody knows how fast it will be. Crucial breakthroughs could happen in days or decades.
-
-
Hmm. Having worked on some of the cutting edge
#Machinelearning techniques I am doubtful it will be within days. Where do you think artificial general intelligence will come from?1 reply 0 retweets 0 likes -
Possibly automation of search for meta learning algorithms.
0 replies 0 retweets 1 like
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.