No smilies, no avatars, no flashing gifs. Just discuss the issues of the day, from last night's telly via football to science or philosophy.
Started by JoolzW on Jan 3, 2019 4:16:14 PM
Free will and sentient AI

As far as I’m aware, the current scientific consensus is that sentience in humans is purely a consequence of our biological building blocks. And given that these building blocks can be broken down into their smallest constituent parts, then one could say that we’re no more than extremely advanced machines. However, if this is really the case, then at some point in the not too distant future, then it’s highly likely that the machines, robots and networks of the future will emerge as sentient beings. Do you believe that this will ever happen? If not, why not? And if they do, what are the ethical implications of this? Should we be sent to prison if we “murdered” our personal robot, for example? Would the robot have a (free) will of its own?

Previous
|
Next
|
Top
|
Bottom
Lento_ - 03 Jan 2019 16:47:19 (#1 of 451)

I tend towards a Turing-ish approach to this of "if we can't tell two things apart then we may as well treat them as being the same". If an AI seems like it is conscious then I'd treat it as if it is.

We are a vast way from that at the moment though, and I doubt we'll be seeing the kinds of AIs you get in sci-fi films any time soon, if ever.

My guess is that we'll develop AIs which become ever better at problem solving, but do not tend towards consciousness. Consciousness is weird and complicated, and may not be necessary in AIs which suit our needs.

widenation - 03 Jan 2019 16:58:48 (#2 of 451)

It's just the spirit of the age.

InternationalVicar - 03 Jan 2019 17:05:19 (#3 of 451)

Like the 1950s were fixated on interplanetary travel, because of a few rockets.

The 60s not so much, despite huge leaps forward.

Hilary - 03 Jan 2019 17:13:51 (#4 of 451)

If a human being cannot tell the difference between a human being and robot AI, this may be because the robot AI has become human-like, but it may equally well be because the human has become robot-like. As human interactions with artificial AI become more frequent, the latter scenario becomes more and more likely.

A third possibility is that the human being attempting to distinguish between robot and human being has herself become insufficiently human to be able to tell the difference (equivalently: sufficiently robotic not to be able to).

Consciousness? "If I didn't have it, I'd miss it. As the general said to the whore, as the whore said to the general...."

slicey - 03 Jan 2019 17:15:16 (#5 of 451)

Microprocessors are a better parallel than rockets as it will radically change pretty much everything over the next decade or two. I get precisely zero sense of consciousness or sentience from how it all works though and can't see how it could ever develop that until people start trying specifically to make it do that, likely with different technology.

InternationalVicar - 03 Jan 2019 17:37:07 (#6 of 451)

I was thinking of the rockets from the Zeitgeisty post #2.

Cultural obsessions of the moment having little bearing on the actual future.

Rockets were important but for geo comms not martians.

AI will change everything so much, while side-stepping the consciousness issue.

Who has time to read the moral concerns wrapped up in the published sermons of the 18th century. Nobody. Too busy buying ironmongery and other industrial products

lammaMia - 03 Jan 2019 17:41:42 (#7 of 451)

https://www.imdb.com/title/tt2180271/?ref_=ttpl_pl
_tt

AdrianNTierney - 03 Jan 2019 18:51:23 (#8 of 451)

Possibly one should remember that both AI and fusion power have been promised "within the next twenty years" ever since the Nineteen Fifties.

And neither one's here yet.

slicey - 03 Jan 2019 18:57:10 (#9 of 451)

Er, AI, very much is here. I work with it every day.

bossab2 - 03 Jan 2019 19:03:46 (#10 of 451)

You wait till Alexa says

"I'm sorry YOUR NAME HERE, I can't do that"

For all we know machines may well already be at slug level consciousness

CarlosFandango - 03 Jan 2019 19:09:05 (#11 of 451)

Competence without comprehension.

Lots of instances of this, from termite mounds to lift systems to vehicle ECUs to bacteria.

cozzer - 03 Jan 2019 19:13:37 (#12 of 451)

But not the presidency of the USA.

Anchorman - 03 Jan 2019 19:20:55 (#13 of 451)

AI, very much is here. I work with it every day.



Yes I'd agree AI is here.

Articial consciousness(AC) isn't though and there seems to be a huge difference between AI and AC.

Infact I'm not sure anyone can adequately describe what consciousness is.

slicey - 03 Jan 2019 19:43:43 (#14 of 451)

Competence without comprehension.

That's really just the sentience argument said in a different way and has the same hard to think about because it gets all self-referential and unknowable problem.

But it also kind of misses the point. We have AIs now that do things that, until very recently, we would have thought would require a human brain to do. And that are producing complex behaviour that is that is not only new to humans but they have discovered themselves and without being told anything about how to do it.

Take those two things together and I think you quickly come to a sort of Turing conclusion that it doesn't matter if it's AI or not because in practical terms it's doing what human brains do only better.

Maybe there is some further level of unknown effects (something quantum say) in a human brain and I think we will meet other barriers (compute power mostly) but what we have now is very compelling as major steps long the way. Unless one gets hung up on consciousness and the sense of self.

slicey - 03 Jan 2019 19:44:16 (#15 of 451)

Infact I'm not sure anyone can adequately describe what consciousness is.

Well, quite.

thisonehasalittlehat - 03 Jan 2019 20:29:27 (#16 of 451)

What do you do with ai?

slicey - 03 Jan 2019 20:34:51 (#17 of 451)

Trade financial markets.

thisonehasalittlehat - 03 Jan 2019 20:36:35 (#18 of 451)

Oh.

slicey - 03 Jan 2019 20:58:19 (#19 of 451)

Well you did ask.

InternationalVicar - 03 Jan 2019 20:59:33 (#20 of 451)

Bit of an open goal that

Previous
|
Next
|
Top
|
Bottom
Check Subscriptions
|
Home » Issues