Artificial Intelligence

[ParagonX]

Where do you think AI will be in 5 years? 10? 25? 50?
How long before we see products designed by AI? (And what kind?)
How long before AI applications become ubiquitous?
How long before AI can mimic the human mind and/or pass it?
What do you hope to gain from AI?

[reejis]

If we make a distinction between "industrial AI" and "Scientific AI",

then wrt Industrial AI:

Where do you think AI will be in 5 years? 10? 25? 50?

In a lot more places. It would be too long winded and boring of me to specify.

How long before we see products designed by AI? (And what kind?)

It's already in the past.
Genetic Algorithms have proven to be mavens of design, for example:
Many chip algorithms have been re-discovered by GAs and a handful of new chip patents have also been produced by them in the last five to seven years.

How long before AI applications become ubiquitous?

Fuzzy logic systems in a variety of household appliances.
Airline ticket booking.
Google search engines and other search engines.
Word processors.
Driving route finders.
Container-packing algorithms.
Diagnostic software across a variety of domains.
Mercedes brake-optimization.
Bose headphone noise-filtering.
chess programs
and on and on and on.

How long before AI can mimic the human mind and/or pass it?

It's both ahead and behind at the moment.

What do you hope to gain from AI?

More efficiency.
Automatization.
More time to play.

Now wrt to Scientific AI:

Where do you think AI will be in 5 years? 10? 25? 50?

Nowhere.

How long before we see products designed by AI? (And what kind?)How long before AI applications become ubiquitous?
How long before AI can mimic the human mind and/or pass it?

Infinite time. We're not going to even recognize AI if it occurs. Because, to derive a criterial set to identify a genuinely aritificial intelligence is tantamount to understanding the underpinings of how our material brains work to produce the conscious. Turing's test is just a joke to me- IMO, you can obtain great behavior, but you'd always debate the consciousness issue. Nobody has done better than Turing, either. There is speculation among AI theorists that the essence of consciousness or maybe genuine intelligence is to be aware of layers of functioning in a self-aware manner, but never see the entirety of the process. But that's just the metaphysics of AI. heh.

What do you hope to gain from AI?

If it happens, that is, then the very understanding by analogy of how my mind works, if that is possible (hah!)

[Quincunx]

Don't be fooled. I think ParagonX is a Turing Machine.

[reejis]

[Bo]

Don't be fooled. I think ParagonX is a Turing Machine.

Damn! He was so close to passing the Test.

Hi Reejis, Quincunx! Long time no see!

My three year old son has made me think a lot more about AI in the last couple of years than otherwise. Now I believe reductionism will have little role to play in AI (probably not new to experts). Fuzzy logic is nothing more than a superficial, clumsy imitation of AI. Neural networks and genetic algorithm may be the best hope we have for now.

But the basic principle of AI must be very simple.

1. Massive storage of multimedia info, and massively parallel (fuzzy) search and retrieval thereof.

2. Proximity calculation, including identity calculation, on the stored info. (How much alike is A to B?)

3. Contextual proximity calculation. Context plays such a pivotal role in human intelligence development. I think it's evident that children learn and understand most things THROUGH context. This ties to my conversation with Reejis during my previous incarnation, on relations. Context is to the subject as categories are to sets. A context is a category of relations between the subject and the rest of the universe. Such relations collectively constitute the definition of the subject in the mind or the AI program. And there's no other meaning of the subject besides its category of relations. In other words, relation is the fundamental construct in AI. Object is a derived construct.

4. Some preprogrammed, ambiguous and generic motivations/goals in BIOS. The desire to survive, to be socially accepted, to command attention, to dominate, etc.

Simple enough.

[ParagonX]

Don't be fooled. I think ParagonX is a Turing Machine.

Oh, darn.
I've failed. :p

My three year old son has made me think a lot more about AI in the last couple of years than
otherwise.

Why's that?

Fuzzy logic is nothing more than a superficial, clumsy imitation of AI.

I don't think it's an imitation of AI (or at least, it shouldn't be), it's something that can be useful to AI.

Some preprogrammed, ambiguous and generic motivations/goals in BIOS. The desire to survive, to be
socially accepted, to command attention, to dominate, etc.

I don't think these should be programmed in. I think they ought to emerge naturally from more fundamental programming.

[reejis]

The desire to survive, to be
socially accepted, to command attention, to dominate, etc.

I don't think these should be programmed in. I think they ought to emerge naturally from more fundamental programming.

I agree, I think these things already show up in simulations as "simulated behavior" that was not intentionally put in, but is a outcome of more basic programed strategies.

This ties to my conversation with Reejis during my previous incarnation, on relations

Who are you? I thought I had that discussion with Bo a year ago.

[Bo]

My three year old son has made me think a lot more about AI in the last couple of years than
otherwise.

Why's that?

Watching a baby grow and learn is very revealing in terms of making you think about the learning and cognitive process. We've been conditioned for so long we forget how we started and got here.

For example, one day I was in the kitchen and he was in the bedroom. I called him. No reponse. I called again. He barged over, stopped abruptly, and yelled "WHAT?" :mgreen:

He had just barely begun to speak. I'm sure he didn't know the meaning of the word "what." But he got the when/why/how perfectly, complete with body language, tone, the whole thing.

How does a baby learn the first things? Massive amount of disparate input and repetition of same/similar input. The brain only needs to store the info and run a pattern recognition algorithm. Then the pattern emerges, and he understands, so to speak.

And the pattern is a collection of contextual relations.

Verbal explanation (abstract illustration) comes much later, after the brain has accumulated certain amount of patterns and abstraction.

Some preprogrammed, ambiguous and generic motivations/goals in BIOS. The desire to survive, to be
socially accepted, to command attention, to dominate, etc.

I don't think these should be programmed in. I think they ought to emerge naturally from more fundamental programming.

Ideally, of course. I didn't want to mention it because at this point I have no idea how to do that. This doesn't mean I know how to implement these motivation now. But I feel I could see some dim light. What kind of underlying algorithm could generate all these basic motivations/desires/drives? I can't even begin to speculate.

And you have to stop somewhere in this emulation, otherwise you end up with the goal of creating a real human. A newborn baby has certain basic motivations. If we could go this deep, that's good enough for me...for now.

[ParagonX]

He had just barely begun to speak. I'm sure he didn't know the meaning of the word "what."

He may very well have known what it meant (at least in that context). I have a couple memories of learning how to speak, and I remember knowing what some words meant even though I hadn't used them. Learning "what?" is a very simple and basic response, so I wouldn't be at all surprised if he did understand.

How does a baby learn the first things? Massive amount of disparate input and repetition of same/similar input. The brain only needs to store the info and run a pattern recognition algorithm. Then the pattern emerges, and he understands, so to speak.

The brain is hard-wired for excellent pattern recognition capabilities. Most other abilities could be said to emerge from this ability, and understanding can come from context.

This doesn't mean I know how to implement these motivation now. But I feel I could see some dim light. What kind of underlying algorithm could generate all these basic motivations/desires/drives? I can't even begin to speculate.

I imagine it would be highly modular, have excellent pattern recognition, be self-referential, and have complex emergent behaviours.

[Bo]

He may very well have known what it meant (at least in that context).

Well, if he knows how to apply it in the proper context, then in the QM observable sense, he understands it. What I meant by "he doesn't understand" is that he had no idea about the spelling, grammar, "meaning" in the dictionary sense -- what we typically think of when we casually say "he understands the word."

I imagine it would be highly modular, have excellent pattern recognition, be self-referential, and have complex emergent behaviours.

How could pattern recognition, in no matter how twisted and winded way, give rise to the desire to survive? Same goes with all the other qualities you mentioned. This is what I meant by "I can't even begin to speculate." Whatever underlie these basic motivations are much deeper than things like pattern recognition.

And it's quite possibly irreducible -- certain basic desires are simply hardcoded in the genes, not results of any underlying mechanism (besides, of course, the pure physical/physiological foundation).

[ParagonX]

Well, if he knows how to apply it in the proper context, then in the QM observable sense, he understands it. What I meant by "he doesn't understand" is that he had no idea about the spelling, grammar, "meaning" in the dictionary sense -- what we typically think of when we casually say "he understands the word."

There's no difference. You don't need to understand spelling and grammar to understand the meaning of a word. Grammar is only necessary when dealing with sentences, and spelling when reading and/or writing.

How could pattern recognition, in no matter how twisted and winded way, give rise to the desire to survive? Same goes with all the other qualities you mentioned. This is what I meant by "I can't even begin to speculate." Whatever underlie these basic motivations are much deeper than things like pattern recognition.

Well, when context is considered, pattern recognition could lead to understanding. That's probably how we start out learning. Understanding death would likely lead to the goal of survival, considering it's undesirable.

And it's quite possibly irreducible -- certain basic desires are simply hardcoded in the genes, not results of any underlying mechanism (besides, of course, the pure physical/physiological foundation).

Well, survival of the genes is hard coded into the genes, but survival of the organism is an emergent behaviour.

[VII I XIII V XX]

AI has many applications and is applied in many areas, usually where either repetitive behaviour is
necessary (a non-biological machine doesn't sleep and its mistakes are our own) or massive calculations
need to be made like in CSP 'Constraint Satisfaction Problem' which could simultaneously involve millions
of inter constrained variables ( which is a simulated task designed by us using heuristics etc, not very
different to a calculator...).

I don't think AI exists which is currently representative of a living concious being because it cannot teach itself ( where it* is a
program and a set of data structures essentially ); so with this said my point is that attention needs to
be paid to the A in AI 'artificial', it's a mimic of intelligence, our best attempts so far are nothing but mimics as far as i
am aware.

Perhaps given a computer that physically evolves and teaches itself could eventually be produced, as to whether
it is concious or not... I think i'm correct in stating the following: Conciousness is an awareness of self, a relationship
between a living entity and its environment or basicly its senses. Which i suppose if you really look at it means to be
concious is to experience one self.

Basicly i expect it'll be very difficult to build a genuinely intelligent mechanical life form and the question of whether it's
concious or not depends upon the definition of conciousness, untill we can all agree on a standardised term it's hard to
deal with the issue.

Thoughts?

[VII I XIII V XX]

Cognitive theoretic model of the Universe

http://www.ctmu.org/

My apologies for going slightly off-topic, but i felt this link
could potentially be relevant to this debate. Have a look it
is a good read, its author is Chris Langan.

Enjoy.

[Aarin]

I'll give it a read but with my short-term memory only once if I learn jack all thats too bad