tech

Rise of the machines: Keep an eye on AI, experts warn

14 Comments

The requested article has expired, and is no longer available. Any related articles, and user comments are shown below.

© 2016 AFP

©2019 GPlusMedia Inc.

14 Comments
Login to comment

Let's hope we won't be needing a real Sarah Connor (!), but it's starting to sounds like we might.

2 ( +2 / -0 )

The game of Go... has more possible board configurations than there are atoms in the Universe.

That's not possible. Someone read some hype and thought it was fact.

9 ( +10 / -1 )

Or matrix, Im not going to be plugged in to go fight those machines!

1 ( +1 / -0 )

its an interesting point in human history we seem to be approaching. AI has the potential to vastly improve the well being of all of us, but also might accidentally turn planet earth into a giant paper clip in space or lead to some other bizarre, unforeseen outcome that destroys humanity if we aren`t super cautious with it.

0 ( +0 / -0 )

Humans actually do not know how big the Universe is so how could we actually know how many atoms there are ?

There are about 10 to the 360 power moves in go or so I hear. Chess only has around 10 to the 21 power moves.

-1 ( +0 / -1 )

According to Wikipedia:

On a 19 x 19 board, there are 2.08168199382... × 10^170 legal positions.

"It has been estimated that the observable universe contains around 10^80 atoms, far fewer than the number of possible legal positions."

0 ( +1 / -1 )

At universities in Europe and elsewhere, students are being advised to seek work in jobs that can't be replaced by robots. I have read two books that delve into this problem ... that is, of robots replacing humans in the workplace. For those entering the workforce as young people ... the AI future can be downright scary ...

I enjoy beating the computer when playing chess, but then I never play at the ultra-high expert level. I tend to play at the lower levels. Love crushing the AI brains ...

0 ( +0 / -0 )

@edojin

Love crushing the AI brains ...

If you beat an AI chess opponent, it's because it let you win.

I'm sorry Dave. I'm going to have to checkmate now.

1 ( +2 / -1 )

“It will be necessary to develop an entirely new discipline of research in order to ensure that increasingly powerful AI systems remain completely under human control… there is a lot of work to do.”

I have total control over the most powerful AI brains! (Pulls out power plug from the outlet)

Seriously people, stop watching "Colossus: The Forbin Project". Ever since 1970 (yes, WAY before "Wargames" or "Terminator"), people have quaked in fear over supercomputers taking over the world. If things get away from the handlers, they'll just pull the plug / trip the breaker / etc.

-2 ( +0 / -2 )

Pulls out power plug from the outlet

But before you do that, they'll have encrypted your data, set booby traps, and left a little note politely suggesting that you switch them back on again if you want to see your cat pictures again.

2 ( +2 / -0 )

If things get away from the handlers, they'll just pull the plug / trip the breaker / etc.

The problem being that any AI worth its salt would foresee that possibility and adopt its strategy accordingly.

AI that we are dealing with today isn`t smart enough to do that, but as it develops the pace of development grows exponentially. Likewise the ability of AI to learn will grow exponentially, something that human brains cannot do.

The fact that people have been worried about this since the 1970s and nothing has happened yet is not an actual substantive argument. In the 1970s we didn`t have computers beating us at Go, nor did we have this thing called the Web which would enable any AI that could learn to access, digest and analyze the entire sum of human knowledge more or less instantaneously and then decide what it wanted to do with that information. Researchers in the field all seem to agree this is something we have to be very concerned about actually happening because they now see a path of development by which it likely will occur. This distinguishes them from people in the 1970s who were mostly thinking in the abstract about what might someday happen.

1 ( +1 / -0 )

"If things get away from the handlers, they'll just pull the plug / trip the breaker / etc." Guessing you've never watched "The Matrix". Sure, it's a fictional movie for now, but self-piloting drones were also science fiction back in the 80's. Technology advancements make self-sufficient AI a distinct possibility.

0 ( +0 / -0 )

The fact that people have been worried about this since the 1970s and nothing has happened yet is not an actual substantive argument.

My point was missed, apparently. Since 1970 the potential has been recognized and nobody is ever going to put an AI in a position where they can dictate terms to humans. There will always be a kill switch.

0 ( +0 / -0 )

Fadamor - yes, except that if we are developing AI that is smarter than us no kill switch we devise is going to be sufficient. no kill switch is fool proof and given the AIs ability to learn it will figure out weaknesses that our human brains aren't even aware exist (much like this one beat us at Go using strategies humans weren't, after centuries of playing the game, aware of).

Not thatthe Go machine can do that, but if we develop AI that can learn exponentially in general terms (not solely confined to the performance of one task) then any safeguards we take would be about as effective as safeguards an ant would use to try and protect its hill from humans.

1 ( +1 / -0 )

Login to leave a comment

Facebook users

Use your Facebook account to login or register with JapanToday. By doing so, you will also receive an email inviting you to receive our news alerts.

Facebook Connect

Login with your JapanToday account

User registration

Articles, Offers & Useful Resources

A mix of what's trending on our other sites