"AI is a dream we shouldn't be having"

[Teallaura]

Robotics expert Noel Sharkey used to be a believer in artificial intelligence, but now thinks AI is a dangerous myth that could lead to a dystopian future of unintelligent, unfeeling robot carers and soldiers. Nic Fleming finds out

Source

I found this really interesting.

[Carrikature]

I can't say I'm surprised. Most of the work on AI, even assuming a computational theory of mind, is currently too far behind the complexity needed to achieve sentience. The pursuit seems to be simulation, not emulation.

As far as AI wanting to take over the world, those ideas have a lot of elements that Sharkey seems to not be aware of. Some of it is an AI seeing humans as a threat and seeking to eliminate the threat. Another is an AI put in place to protect us and implementing protections we rail against.

I think there's a more important question for humans to think about when creating robots and AI to replace us. What will we be doing with our time when robots are performing the work?

[Jedidiah]

The goal is definitely simulation. There is no actual explanation for sentience in existence.

[KingsGambit]

I think there's a more important question for humans to think about when creating robots and AI to replace us. What will we be doing with our time when robots are performing the work?

This is a sobering thought. And yet... at least since the reforms at the beginning of the Industrial Revolution, it seems that technology doesn't always reduce the amount of work we do; it simply justifies us having even more on our plate now that we have the capacity to handle it.

[Carrikature]

There is no actual explanation for sentience in existence.

That's not really accurate.

[Sparko]

[Carrikature]

This is a sobering thought. And yet... at least since the reforms at the beginning of the Industrial Revolution, it seems that technology doesn't always reduce the amount of work we do; it simply justifies us having even more on our plate now that we have the capacity to handle it.

There are different kinds of technology, in my opinion, but it depends on how you look at it. I can send a mass email near instantly instead of hand-writing a dozen different letters and sending them to be carried on horseback or ship and waiting weeks or months for a response. Consider what used to be required to produce engineered drawings: hand-drawn with stencils on special paper that didn't let you erase well if at all. I would call that a massive reduction in the amount of work that I do.

You're right, though, that we've always found more to do. I'm not sure how well that holds when/if we develop the ability to do complete tasks with either no input or just programming. Imagine an automated, self-diagnosing entity that communicates with an automated overseer for repairs and deployment based on measurements of various conditions. I suspect there will always be people required, but the quantity needed drops significantly.

[Teallaura]

I thought of that one, too...

[Sparko]

There are different kinds of technology, in my opinion, but it depends on how you look at it. I can send a mass email near instantly instead of hand-writing a dozen different letters and sending them to be carried on horseback or ship and waiting weeks or months for a response. Consider what used to be required to produce engineered drawings: hand-drawn with stencils on special paper that didn't let you erase well if at all. I would call that a massive reduction in the amount of work that I do.

You're right, though, that we've always found more to do. I'm not sure how well that holds when/if we develop the ability to do complete tasks with either no input or just programming. Imagine an automated, self-diagnosing entity that communicates with an automated overseer for repairs and deployment based on measurements of various conditions. I suspect there will always be people required, but the quantity needed drops significantly.

I think you are confusing automation with artificial intelligence.

The OP article is saying that AI is really good at faking intelligence but when it comes down to it, it is just a brute force method of digging through an expert system database to come up with answers that sound like it is thinking. A parlor trick basically. We are not any closer to a true sentient artificial mind now than we were 100 years ago. Just better at faking it.

[Teallaura]

There are different kinds of technology, in my opinion, but it depends on how you look at it. I can send a mass email near instantly instead of hand-writing a dozen different letters and sending them to be carried on horseback or ship and waiting weeks or months for a response. Consider what used to be required to produce engineered drawings: hand-drawn with stencils on special paper that didn't let you erase well if at all. I would call that a massive reduction in the amount of work that I do.

You're right, though, that we've always found more to do. I'm not sure how well that holds when/if we develop the ability to do complete tasks with either no input or just programming. Imagine an automated, self-diagnosing entity that communicates with an automated overseer for repairs and deployment based on measurements of various conditions. I suspect there will always be people required, but the quantity needed drops significantly.

Stencils could be corrected to a degree (I still like the smell of correction fluid. Reminds me of my Mom...) - but there was a reason draftsmen were well paid. Errors could easily destroy days of work on blueprints.

[Sparko]

Stencils could be corrected to a degree (I still like the smell of correction fluid. Reminds me of my Mom...) - but there was a reason draftsmen were well paid. Errors could easily destroy days of work on blueprints.

I started out as an electrical draftsman, using a drafting machine (a square and ruler mounted to a drafting table) and templates.

In fact, my table looked almost exactly like this:


We drew on vellum which was fairly easy to erase mistakes on. When we were done, we copied the drawings onto blueprints (a type of copy) to make them permanent and create working drawings.

and no I wasn't well paid.

I even helped to bring in and set up Autocad when it first became available. It saved some time and allowed us to store pieces to be reused, and had built in symbols instead of having to use templates.

[Teallaura]

I started out as an electrical draftsman, using a drafting machine (a square and ruler mounted to a drafting table) and templates.

In fact, my table looked almost exactly like this:


We drew on vellum which was fairly easy to erase mistakes on. When we were done, we copied the drawings onto blueprints (a type of copy) to make them permanent and create working drawings.

and no I wasn't well paid.

I even helped to bring in and set up Autocad when it first became available. It saved some time and allowed us to store pieces to be reused, and had built in symbols instead of having to use templates.

We had a mimeograph when I was a girl (my parents owned a kindergarten for awhile). I seem to recall draftsmen who dealt with stencils being well - or at least better - paid but I could be mistaken. I know stencils could be corrected but the corrections didn't really look good and didn't print nearly as well as non-corrected areas.

[Carrikature]

I think you are confusing automation with artificial intelligence.

You would be wrong, then. I've spent enough time on the subject not to make that mistake.

The OP article is saying that AI is really good at faking intelligence but when it comes down to it, it is just a brute force method of digging through an expert system database to come up with answers that sound like it is thinking. A parlor trick basically. We are not any closer to a true sentient artificial mind now than we were 100 years ago. Just better at faking it.

I understood what the article is saying. I also know that it's an incomplete picture of what really is taking place these days. I've seen robots that are actively learning. Yes, we have really strong systems that are just database-lookup programs. In truth, part of intelligence is database lookup. It's more than that, of course, but you don't have inference and pattern recognition without remembrance of previous encounters.

[Carrikature]

Stencils could be corrected to a degree (I still like the smell of correction fluid. Reminds me of my Mom...) - but there was a reason draftsmen were well paid. Errors could easily destroy days of work on blueprints.

I've heard lots of horror stories about thin paper ripping with too much pressure.

[Teallaura]

I've heard lots of horror stories about thin paper ripping with too much pressure.

Oh yeah - and 'too much' didn't amount to much at all. Then there were those horrible moments when the stencil caught on something and got ripped to shreds in the mimeograph...