Go is a very complex game with simple rules. This game is one of
the many things that keep me humble. I play go--badly. I've played
enough decent players to know how bad I am. To make it interesting
for them semi-serious players have to give me at least a 2 or 3
stone handicap. IOW I start the game by playing extra stones. If we
hit the right handicap number I should win half the time. The
people I give a handicap to are seriously weak.
Go is significantly more complex than chess. So it was a bit of a
surprise when the machines started beating skilled players. This
defeat and some of the other behavior we're seeing now that AI is
out in the wild clearly demonstrates that AI isn't nearly as smart
as the sales people would like us to believe.
"The discovery of a weakness in some of the most advanced
Go-playing machines points to a fundamental flaw in the deep
learning systems that underpin today’s most advanced AI, said
Stuart Russell, a computer science professor at the University of
California, Berkeley.
The systems can “understand†only specific situations they have been exposed to in the past and are unable to generalize in a way
that humans find easy, he added.
“It shows once again we’ve been far too hasty to ascribe superhuman levels of intelligence to machines,†Russell said."
https://arstechnica.com/information-technology/2023/02/man-beats-machine-at-go-in-human-victory-over-ai/?comments=1&comments-page=1
Technobarbarian wrote:
Go is a very complex game with simple rules. This game is one of
the many things that keep me humble. I play go--badly. I've played
enough decent players to know how bad I am. To make it interesting
for them semi-serious players have to give me at least a 2 or 3
stone handicap. IOW I start the game by playing extra stones. If we
hit the right handicap number I should win half the time. The
people I give a handicap to are seriously weak.
Go is significantly more complex than chess. So it was a bit of a
surprise when the machines started beating skilled players. This
defeat and some of the other behavior we're seeing now that AI is
out in the wild clearly demonstrates that AI isn't nearly as smart
as the sales people would like us to believe.
"The discovery of a weakness in some of the most advanced
Go-playing machines points to a fundamental flaw in the deep
learning systems that underpin today’s most advanced AI, said
Stuart Russell, a computer science professor at the University of California, Berkeley.
The systems can “understand†only specific situations they have been exposed to in the past and are unable to generalize in a way--------------------------------------------------------------------
that humans find easy, he added.
The tactics that put a human back on top on the Go board were
suggested by a computer program that had probed the AI systems looking
for weaknesses. The suggested plan was then ruthlessly delivered by
Pelrine.
...
The decisive victory, albeit with the help of tactics suggested by a computer
...
-------------------------------------------------------------------- "albeit"? LOL. I allege that the available evidence provided in the
article strongly suggests that the help was "crucial" to the decisive victory.
“It shows once again we’ve been far too hasty to ascribe superhuman levels of intelligence to machines,†Russell said."
Because the "discovery of a weakness" and the use of tactics suggested
by Computer B are what helped a human to beat Computer A, I allege
that they're being "far too hasty" in underestimating the
"intelligence" - and in particular, the potential intelligence - of machines.
I further allege that if we continue to literally wallow in our hubris
in this context, there are going to problems - maybe even Problems - appearing on the table at the end of the day going forward.
https://arstechnica.com/information-technology/2023/02/man-beats-machine-at-go-in-human-victory-over-ai/?comments=1&comments-page=1
--
bill
Theory don't mean squat if it don't work.
On Sunday, February 19, 2023 at 10:57:37 AM UTC-8, bfh wrote:The tactics that put a human back on top on the Go board were
Technobarbarian wrote:
--------------------------------------------------------------------
Go is a very complex game with simple rules. This game is one
of the many things that keep me humble. I play go--badly. I've
played enough decent players to know how bad I am. To make it
interesting for them semi-serious players have to give me at
least a 2 or 3 stone handicap. IOW I start the game by playing
extra stones. If we hit the right handicap number I should win
half the time. The people I give a handicap to are seriously
weak.
Go is significantly more complex than chess. So it was a bit of
a surprise when the machines started beating skilled players.
This defeat and some of the other behavior we're seeing now
that AI is out in the wild clearly demonstrates that AI isn't
nearly as smart as the sales people would like us to believe.
"The discovery of a weakness in some of the most advanced
Go-playing machines points to a fundamental flaw in the deep
learning systems that underpin today’s most advanced AI, >>> said Stuart Russell, a computer science professor at the
University of California, Berkeley.
The systems can “understand†only specific situations
they have been exposed to in the past and are unable to
generalize in a way that humans find easy, he added.
"albeit"? LOL. I allege that the available evidence provided in thesuggested by a computer program that had probed the AI systems
looking for weaknesses. The suggested plan was then ruthlessly
delivered by Pelrine. ... The decisive victory, albeit with the
help of tactics suggested by a computer ...
--------------------------------------------------------------------
--article strongly suggests that the help was "crucial" to the
decisive victory.
“It shows once again we’ve been far too hasty to
ascribe superhuman levels of intelligence to machines,â€
Russell said."
Because the "discovery of a weakness" and the use of tactics
suggested by Computer B are what helped a human to beat Computer
A, I allege that they're being "far too hasty" in underestimating
the "intelligence" - and in particular, the potential
intelligence - of machines.
I further allege that if we continue to literally wallow in our
hubris in this context, there are going to problems - maybe even
Problems - appearing on the table at the end of the day going
forward.
https://arstechnica.com/information-technology/2023/02/man-beats-machine-at-go-in-human-victory-over-ai/?comments=1&comments-page=1
bill Theory don't mean squat if it don't work.
heh, Board games aren't all that popular these days. If this had
been a video game gamers would have stumbled on this brilliantly
stupid strategy, exploited it and shared it years ago. And it won't
be an easy hack to fix because the software depends on its own
hacks. There are so many possibilities that the machine can't
consider all of them. It has to look at the paths that are
mathematically most likely to produce the best results. If it has
to start worrying about this hack, it has to pay less attention to
other possibilities.
Feb 18, 2023,11:40am EST
Microsoft announced it was placing new limits on its Bing chatbot
following a week of users reporting some extremely disturbing
conversations with the new AI tool. How disturbing? The chatbot
expressed a desire to steal nuclear access codes and told one reporter
it loved him. Repeatedly. ------------------------------------------------------- https://www.forbes.com/sites/mattnovak/2023/02/18/microsoft-puts-new-limits-on-bings-ai-chatbot-after-it-expressed-desire-to-steal-nuclear-secrets/?sh=57920373685c
damthing's a pervert potential terrorist. I'll bet it's probably also identity-conflicted - can't decide if itself is Android, Windows, or
Apple. I recommend a PC Matic checkup no later than at the end of the
day going forward.
--
bill
Theory don't mean squat if it don't work.
On Sunday, February 19, 2023 at 5:46:06 PM UTC-8, bfh wrote:
Feb 18, 2023,11:40am EST
Microsoft announced it was placing new limits on its Bing chatbot
following a week of users reporting some extremely disturbing
conversations with the new AI tool. How disturbing? The chatbot
expressed a desire to steal nuclear access codes and told one reporter
it loved him. Repeatedly.
-------------------------------------------------------
https://www.forbes.com/sites/mattnovak/2023/02/18/microsoft-puts-new-limits-on-bings-ai-chatbot-after-it-expressed-desire-to-steal-nuclear-secrets/?sh=57920373685c
damthing's a pervert potential terrorist. I'll bet it's probably also
identity-conflicted - can't decide if itself is Android, Windows, or
Apple. I recommend a PC Matic checkup no later than at the end of the
day going forward.
--
bill
Theory don't mean squat if it don't work.
""My Saturday fun project: using AI, every US president as a Pixar character.""
https://www.democraticunderground.com/100217661473
TB
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 297 |
Nodes: | 16 (2 / 14) |
Uptime: | 96:31:05 |
Calls: | 6,659 |
Calls today: | 1 |
Files: | 12,206 |
Messages: | 5,334,482 |