One Big Question: Why is artificial intelligence still kind of dumb?

Once the domain of science-fiction authors and script writers, artificial intelligence is steadily marching into the real world. Recently we’ve seen the technology do everything from reading lips better than the experts to playing in a poker tournament and trouncing its human competition.

Will robots and AI be emotionally intelligent?

Will their emotions be analog like ours?

How many emotions will robots have?

Will they experience multiple emotions at the same time?

Will robot emotions be limited?

Will robots emotions mimic humans or be totally different?

Tau.Neutrino said:

Will robots and AI be emotionally intelligent?

Will their emotions be analog like ours?

How many emotions will robots have?

Will they experience multiple emotions at the same time?

Will robot emotions be limited?

Will robots emotions mimic humans or be totally different?

yes. no need for all though.
yes. save confusion
100. or so.
yes. if they are analogue to ours…
depends on the role.
they’ll be analogue to ours.

In the past I would have said that artificial intelligence is so dumb because the researchers in the field in artificial have no natural intelligence.

But let’s look deeper into it than that.

One thing almost totally missing from the field of artificial intelligence is the ability to randomly forget. In order to learn new things, an organic being has to forget old ones. But workers in the field of artificial intelligence value computer memory too highly to allow the computer to randomly monkey around with it.

Another thought I had ages ago is that intelligence is more than a meme, it can be thought of as a separate life form that parasitises human beings. So we can posit that artificial intelligence is so dumb because either it hasn’t learnt to parasitise hardware well enough yet, or we can posit that there are two types of intelligence ie. book smart vs street smart. Book smart intelligence inhabits Google, street smart intelligence only inhabits carbon-based life.

On another level, government agencies don’t want artificial intelligence to be smart – they don’t want Mr Joe Public to have access to military-grade artificial intelligence such as voice and text recognition and smart guidance systems.

AI research is divided into subfields that have no overlap. For instance, an “expert system” is just a book with hyperlinks. As a result, it can be claimed that there really is no such thing as AI, just a loose collection of unrelated projects.

Robot emotions.

Robot emotions are easy to program, we’ve seen them in R2D2, in Wall-E, in Short Circuit, and in the Sirius Cybernetics Corporatioo.

But the real question is “why?”

In R2D2, emotions have survival value, both for the individual (eg. fear) and for the collective (eg. bravery).

Another purpose of emotions is interactions with people. The Sirius Cybernetics Corporation defines a robot as “your plastic pal who’s fun to be with”.

In Wall-E, emotions develop in robots … Wall-E is an interesting case. The emotions there have nothing to do with either survival or interactions with people. Instead, there is a balance between rational behaviour and irrational behaviour, the yin and yang. Do you want to program robots to be irrational? Why not?

mollwollfumble said:

Robot emotions.

Robot emotions are easy to program, we’ve seen them in R2D2, in Wall-E, in Short Circuit, and in the Sirius Cybernetics Corporatioo.

But the real question is “why?”

In R2D2, emotions have survival value, both for the individual (eg. fear) and for the collective (eg. bravery).

Another purpose of emotions is interactions with people. The Sirius Cybernetics Corporation defines a robot as “your plastic pal who’s fun to be with”.

In Wall-E, emotions develop in robots … Wall-E is an interesting case. The emotions there have nothing to do with either survival or interactions with people. Instead, there is a balance between rational behaviour and irrational behaviour, the yin and yang. Do you want to program robots to be irrational? Why not?

Are these emotions (feelings), or programed parameters?

PermeateFree said:

mollwollfumble said:

Robot emotions.

Robot emotions are easy to program, we’ve seen them in R2D2, in Wall-E, in Short Circuit, and in the Sirius Cybernetics Corporatioo.

But the real question is “why?”

In R2D2, emotions have survival value, both for the individual (eg. fear) and for the collective (eg. bravery).

Another purpose of emotions is interactions with people. The Sirius Cybernetics Corporation defines a robot as “your plastic pal who’s fun to be with”.

In Wall-E, emotions develop in robots … Wall-E is an interesting case. The emotions there have nothing to do with either survival or interactions with people. Instead, there is a balance between rational behaviour and irrational behaviour, the yin and yang. Do you want to program robots to be irrational? Why not?

Are these emotions (feelings), or programed parameters?

In my mind, the two are equivalent. In an earlier CN thread about emotions of robots, I claimed that the difference between a real emotion and a faked emotion is the same as that between real art and faked art – the fidelity of the fake. If you fake being happy then it won’t make you happy, but if you fake being happy with perfect fidelity then you are happy.

because it’s more generally elusive in nature

transition said:

because it’s more generally elusive in nature

Because it stands to reason that if there is artificial intelligence then there must also be artificial stupidity.

Tamb said:

transition said:

because it’s more generally elusive in nature

Because it stands to reason that if there is artificial intelligence then there must also be artificial stupidity.

That’s an excellent point, Tamb. If you constrain an AI system so as to remove any chance of artificial stupidity, then you also remove any chance of artificial intelligence.

Tamb said:

transition said:

because it’s more generally elusive in nature

Because it stands to reason that if there is artificial intelligence then there must also be artificial stupidity.

i’ve often thought like you know the geniuses generalize intelligence to be something like …the resources brought to bear in overcoming obstacles in achieving goals (sort of thing, whatever), but of the objectives/goals, what of those.

what are intelligent objectives, of any bigger picture.

of course this is more the territory of philosophy, politics too. The workings of the aether.

If they ever develop an AI that is truly conscious, it might require very fine tuning to ensure that it’s sane.

Bubblecar said:

If they ever develop an AI that is truly conscious, it might require very fine tuning to ensure that it’s sane.

There was something wrong with Data’s brother.

> i’ve often thought like you know the geniuses generalize intelligence to be something like …the resources brought to bear in overcoming obstacles in achieving goals (sort of thing, whatever), but of the objectives/goals, what of those. What are intelligent objectives, of any bigger picture.

The best definition of intelligence that I know is “Intelligence is knowing what do do when you don’t know what to do”. Try programming that into an AI system.

Bubblecar said:

If they ever develop an AI that is truly conscious, it might require very fine tuning to ensure that it’s sane.

I’m really enjoying this conversation.

> what do do

what to do

on the subject of the the experience of organic intelligence, of animals, and this primate mammal vehicle I inhabit, a few things strike me as important, somewhat neglected maybe.

the area of skin humans have is a large contributor to the sensory array, and it’s not (all, or mostly) passive.

the other thing is the experience of blood circulating through our body (and brain), oxygenated blood. Being fueled too.

much as consciousness (or internal mental states) give a nice (somewhat insulated) inner environment, the experience is still of something fleshy.

AI doesn’t have the experience of fleshy. That sort of homeostasis.

mollwollfumble said:

The best definition of intelligence that I know is “Intelligence is knowing what to do when you don’t know what to do”. Try programming that into an AI system.

Intelligence mindmap, so you can program it into an AI system.

mollwollfumble said:

mollwollfumble said:

The best definition of intelligence that I know is “Intelligence is knowing what to do when you don’t know what to do”. Try programming that into an AI system.

Intelligence mindmap, so you can program it into an AI system.

A few more items added to the mindmap for intelligence:

Do the part you know how to do
Sleep on it
Trial and error
Remove obstacle

I wonder if there’s a similar mindmap for stupidity? Options would include, perhaps:

Commit to the wrong choice
Panic
Give up (note that this is different to “run”)
Corrupt database
Illogic
Failure to learn
Get lost in the complexity
Blindness to alternatives
Failure to recognise the true solution
Ridiculous objective
Procrastination
Deceit
Cruelty

There you go, now program all those into an AI.

Will artificial intelligence kill us all?

No – Robots Won’t Try To Kill Us, Says Stanford’s 100-Year Study Of AI

Yes – Jay Tuck, US defence expert

Ian said:

Will artificial intelligence kill us all?

No – Robots Won’t Try To Kill Us, Says Stanford’s 100-Year Study Of AI

Yes – Jay Tuck, US defence expert

Didn’t watch the JT video, but did read a comment or two:

The speaker is a journalist, not an expert on artificial intelligence. He seems to cast a wide net, labeling any sophisticated software AI. He repeatedly claims software operates as a “brain”, revealing that he understands brain function as little as he understands software. He is not worthy of any version of TED, and is rather more suited for a starring role in a “reality” television show.

The Rev Dodgson said:

Ian said:

Will artificial intelligence kill us all?

No – Robots Won’t Try To Kill Us, Says Stanford’s 100-Year Study Of AI

Yes – Jay Tuck, US defence expert

Didn’t watch the JT video, but did read a comment or two:

The speaker is a journalist, not an expert on artificial intelligence. He seems to cast a wide net, labeling any sophisticated software AI. He repeatedly claims software operates as a “brain”, revealing that he understands brain function as little as he understands software. He is not worthy of any version of TED, and is rather more suited for a starring role in a “reality” television show.

Can AI be judgmental?

The Rev Dodgson said:

Ian said:

Will artificial intelligence kill us all?

No – Robots Won’t Try To Kill Us, Says Stanford’s 100-Year Study Of AI

Yes – Jay Tuck, US defence expert

Didn’t watch the JT video, but did read a comment or two:

The speaker is a journalist, not an expert on artificial intelligence. He seems to cast a wide net, labeling any sophisticated software AI. He repeatedly claims software operates as a “brain”, revealing that he understands brain function as little as he understands software. He is not worthy of any version of TED, and is rather more suited for a starring role in a “reality” television show.

That’s a pretty strong conclusion for not having watched the video but I generally agree with you. He points out that there was demonstration military robot that was on the brink of shooting up the audience until tackled by a soldier. He says that big data needs AI and vice verse, and the expanding internet and internet of things will give AI the ability to grow exponentially (I agree with). But the point where he makes the leap to “it will turn around and kill us” is not well made IMO.

I can’t comprehend what does happen to this point however.. there are plenty more YouTubes :)

>Because it stands to reason that if there is artificial intelligence then there must also be artificial stupidity.

I tend to see things as mechanisms. Much of mental activity involves inhibition (inhibitory mechanisms), or restraints, expressed so (or not) in what’s done/behaviours. Even this paragraph is mostly a work of restraint.

Where does stupidity come from? I’m lost for a non-pejorative sense.

Mechanisms are more often than not dumb, one can however attribute them design intelligence. A trick, whatever does whatever.

Nature happened upon many tricks.

It’s tempting to think intelligence is going to be a busy thing. Out there you know, busy, a happener, a fixer of unhappener, and proliferating. And it’s going to venture everywhere it can.

Or, in fact does it do not much of all it could do, sits on its hands? Minds its own business. Perhaps it sleeps a lot. Maybe it’d avoid consciousness.

Which brings me to the question of what happens when something inhabits the territory of (soft) aversions to self-awareness. A weird proposition.

> Where does stupidity come from? I’m lost for a non-pejorative sense.

See mind map of stupidity listed above.

Some people fake emotions give all the right responses to various scenarios and most people buy it probably fairly easy to set up algorithms to mimic this, people often ask tokenistic questions regarding how to do you feel and get a tokenistic response.

A) we don’t know how intelligence works

B) “real” intelligence is also pretty dumb

Dropbear said:

A) we don’t know how intelligence works

B) “real” intelligence is also pretty dumb

Intelligence is figuring out how intelligence works. Oh wait, you’ve just said that.