Thoughts on the coming age. A first of two essays.
12 thoughts on “Artificial Intelligence”
Boy, the comments on that one were grade-school level. I wish Wretchard had pointed them to Bostrum’s book or something.
Fernandez is one of the most thoughtful writers around these days.
And never many comments on his tweets, its crazy.
“Put simply, they’re looking to ‘dim the sun,’” reports WCCO. With that, environmentalism has completed its journey from a movement to leave nature alone to a project to actively terraform the planet.”
This is a great documentary on the subject. Well, not great but amazing in its prescience.
“When “trust the science” is eventually discarded in favor of “trust the AI,” something will still be standing behind the curtain: the oracle.”
A bit like everything is blamed on the algorithm rather than the people who control the algorithm. It worked for Comey and Hillary when they claimed they did some computer stuff to examine all the emails on a single Saturday.
Never been a big fan of A.I. as it currently seems to be defined.
I am a big fan of smart systems. Programs that are really, really good at what they do, but don’t bother with much else. Those I’m okay with.
A.I.s, not so much. It’s not that I don’t trust them, it’s that I trust them to do things that may or may not be in humanity’s favor.
Google’s Lens app has been useful. They do an ok job trying to stuff adds into the results.
“It’s not that I don’t trust them, it’s that I trust them to do things that may or may not be in humanity’s favor.”
Biggest problem/issue with AI is that it would become very quickly better than we are at just about everything. From struggling to beat an average “go” player to being able to beat the best to being able to easily beat the top 5000 in the world simultaneously in a very short period of time. Imagine when it can simultaneously fly thousands of commercial airliners at the same time or perform 10’s of thousands of computer controlled robotic surgery at the same time etc. In no time at all we we all be largely superfluous.
Elon Musk’s Last Warning 2022 – “I Tried To Warn You The Last Few Years” (BEFORE IT’S TOO LATE!!!)
And it could easily manipulate us on mass/social media…”read” our responses facial expressions/body language expertly constantly refining/tailoring its message to be more persuasive. Imagine say the Chinese/Russians leadership developing said AI; in no time at all it would not only learn how to expertly manipulate the populace (the desired effect) but the leadership as well. Your average octogenarian leader type would likely be very easy to control by said AI; he (the leader) would think he was still in control but not really.
Well I think AI could be useful as lawyer or politician.
Mysteriously, Arnold Schwarzenegger and Linda Hamilton are murdered one evening. The police AI quickly finds the perpetrators, who are then mysteriously killed in a runaway elevator accident in the police precinct. The perps are found to have received large payments, but after tracing the payments through ten thousand shell corporations, the banking AI throws up its electronic hands and apologizes, saying it is simply not capable enough, but it is willing to design its replacement for us.
Evolution so far has served as the only reliable oracle. That is unlikely to stop being true with AI’s. Self-driving has gone largely as expected by anyone not smart enough to believe things that can’t be true – an endless recursion loop of 9’s, each growing exponentially harder than the last. That is but a small subset of strong AI. There are thermodynamic limits, and when approaching those limits gray goo starts to look a lot like green goo. Of course, green goo can still be dangerous.
The oracle problem is considered one of the great unsolved problems in computer science, but that undersells it. The inability to predict the future in a formally complex system seems to be one of, if not the, fundamental law of physics. Stephen Wolfram’s recent work seems to suggest irreduceable computation may underly physics itself.
Boy, the comments on that one were grade-school level. I wish Wretchard had pointed them to Bostrum’s book or something.
Fernandez is one of the most thoughtful writers around these days.
And never many comments on his tweets, its crazy.
“Put simply, they’re looking to ‘dim the sun,’” reports WCCO. With that, environmentalism has completed its journey from a movement to leave nature alone to a project to actively terraform the planet.”
This is a great documentary on the subject. Well, not great but amazing in its prescience.
“When “trust the science” is eventually discarded in favor of “trust the AI,” something will still be standing behind the curtain: the oracle.”
A bit like everything is blamed on the algorithm rather than the people who control the algorithm. It worked for Comey and Hillary when they claimed they did some computer stuff to examine all the emails on a single Saturday.
Never been a big fan of A.I. as it currently seems to be defined.
I am a big fan of smart systems. Programs that are really, really good at what they do, but don’t bother with much else. Those I’m okay with.
A.I.s, not so much. It’s not that I don’t trust them, it’s that I trust them to do things that may or may not be in humanity’s favor.
Google’s Lens app has been useful. They do an ok job trying to stuff adds into the results.
“It’s not that I don’t trust them, it’s that I trust them to do things that may or may not be in humanity’s favor.”
Biggest problem/issue with AI is that it would become very quickly better than we are at just about everything. From struggling to beat an average “go” player to being able to beat the best to being able to easily beat the top 5000 in the world simultaneously in a very short period of time. Imagine when it can simultaneously fly thousands of commercial airliners at the same time or perform 10’s of thousands of computer controlled robotic surgery at the same time etc. In no time at all we we all be largely superfluous.
Elon Musk’s Last Warning 2022 – “I Tried To Warn You The Last Few Years” (BEFORE IT’S TOO LATE!!!)
https://www.youtube.com/watch?v=UI7xZDo0LCE
Cheaper just to buy off the politicians?
And it could easily manipulate us on mass/social media…”read” our responses facial expressions/body language expertly constantly refining/tailoring its message to be more persuasive. Imagine say the Chinese/Russians leadership developing said AI; in no time at all it would not only learn how to expertly manipulate the populace (the desired effect) but the leadership as well. Your average octogenarian leader type would likely be very easy to control by said AI; he (the leader) would think he was still in control but not really.
Well I think AI could be useful as lawyer or politician.
Mysteriously, Arnold Schwarzenegger and Linda Hamilton are murdered one evening. The police AI quickly finds the perpetrators, who are then mysteriously killed in a runaway elevator accident in the police precinct. The perps are found to have received large payments, but after tracing the payments through ten thousand shell corporations, the banking AI throws up its electronic hands and apologizes, saying it is simply not capable enough, but it is willing to design its replacement for us.
Evolution so far has served as the only reliable oracle. That is unlikely to stop being true with AI’s. Self-driving has gone largely as expected by anyone not smart enough to believe things that can’t be true – an endless recursion loop of 9’s, each growing exponentially harder than the last. That is but a small subset of strong AI. There are thermodynamic limits, and when approaching those limits gray goo starts to look a lot like green goo. Of course, green goo can still be dangerous.
The oracle problem is considered one of the great unsolved problems in computer science, but that undersells it. The inability to predict the future in a formally complex system seems to be one of, if not the, fundamental law of physics. Stephen Wolfram’s recent work seems to suggest irreduceable computation may underly physics itself.