Artificial intelligence

[Dave B]

I thought there was already a thread looking at this, seems not, do . . .

[Alan H]

Fascinating. What is clear is that it must be true that huge leaps in progress will be made in the coming years and decades. Where will that take us? No idea!

[Dave B]

Fascinating. What is clear is that it must be true that huge leaps in progress will be made in the coming years and decades. Where will that take us? No idea!

Yeah, when you realise just how many big and small players are in the AI business . . .

[Compassionist]

Great video Dave. Thanks for sharing.

[Ken H]

Nice video, Dave.

Philip K. Dick lives (as a robot).

[Dave B]

Nice video, Dave.

Philip K. Dick lives (as a robot).

And does he dream of electric sheep one wonders.

[Dave B]

Yes, androids do dream of electric sheep

But ‘The most powerful programs in the world are no more human than a shovel is, or a nuclear submarine.’

[Dave B]

Microsoft deletes 'teen girl' AI after it became a Hitler-loving sex robot within 24 hours
Such is the danger of allowing any "entity" to "learn" from its "peers" without any form of "supervision".
(Some loaded words there, chosen for brevity!)

[Alan H]

Microsoft deletes 'teen girl' AI after it became a Hitler-loving sex robot within 24 hours
Such is the danger of allowing any "entity" to "learn" from its "peers" without any form of "supervision".
(Some loaded words there, chosen for brevity!)

What I don't understand is why they let it go on after it - rather obviously - went off the rails after being taunted.

[Dave B]

Microsoft deletes 'teen girl' AI after it became a Hitler-loving sex robot within 24 hours
Such is the danger of allowing any "entity" to "learn" from its "peers" without any form of "supervision".
(Some loaded words there, chosen for brevity!)

What I don't understand is why they let it go on after it - rather obviously - went off the rails after being taunted.

Dunno, but I would bet that any team members who had a psychological or sociological background would have wanted this project left to run. Lots of potential to learn about "learning systems", those that learn and modify their behaviour according to their experience/environment.

In this case it seemed to act much like an abused or neglected child with a temper problem would. Perhaps this will inform those who design future systems?

But do you wait until such a system can be "reasoned with" and led, as one would a human learner, or do you act like a represive parent, ban certain words and subjects and, metaphorically, wash the AI's mouth out with soap? Can an AI learn through "punishment" or does it have to be "restrained" by edict?

One needs a new paradigm of language to even discuss this subject!

Later: perhaps the questions I asked above could be rephrased in terms of; "What is intelligence? Can it be wired in (as ours is in our genes)? Can it be "imposed"? Can it be learned?

How it is applied is a whole new matter, in this case it was, sort of, a an unemotional "copy-cat" response. Assuming AIs cannot, as yet, have emotions - can they express emotional analogues?

Glad I am not an ethicist, or ethical psychologist, involved in this research!