The death of AI (yet again)

Here is an interesting article predicting the unravelling of the Artificial Intelligence (AI) myth in 2017:

http://www.theregister.co.uk/2017/01/02/ai_was_the_fake_news_of_2016/

(Be sure to read the comments at the end of the last page)

I think the concept of AI resurfaces from time to time because there are so many people who are convinced that consciousness just has to be a materialistic phenomenon, that AI has to work! It isn't, so it doesn't!

David
 
David,

Is there a technical definition of AI? Who determines how AI is defined?

Is there a difference between artificial intelligence and artificial consciousness?

Why isn't a program that can perform better than a human (like a chess program, a Jeopardy program, or a medical diagnosis program) AI?
 
Last edited:
David,

Is there a technical definition of AI? Who determines how AI is defined?

Is there a difference between artificial intelligence and artificial consciousness?
I think that you have put your finger on two of the real problems with AI - it isn't clearly defined, yet from time to time it is promoted as something all powerful.
Why isn't a program that can perform better than a human (like a chess program, a Jeopardy program, or a medical diagnosis program) AI?
This is also rather vague, but programs that have a very narrow scope (e.g. just solve a specific set of differential equations), or which use purely brute force methods, are not called AI, because they don't seem human enough. Indeed in the early days, computers that could (gasp) sum up thousands of numbers in a second, were referred to as "giant electronic brains" - which sort of implies the AI concept.

As my link suggested, AI has always been more about what it implied than what it delivered! It implies that consciousness is equivalent to computation, although people avoid using the term "Artificial Consciousness" because I think they know issues related to the Hard Problem will arise if they use that term.

In its early days, Joseph Weizenbaum developed a little program which engaged in a teletype dialogue with people. The program performed as a psychiatrist, and it fooled a lot of people into thinking it really understood mentality, and could help them with problems without the embarrassment of talking to another human! He then exposed the fact that it was based on a set of very simple tricks. For example, if the user typed a sentence that contained the word "Mother" (presumably including various synonyms) it would reply,

"Tell me more about your mother."

https://en.wikipedia.org/wiki/ELIZA

I think AI is fascinating, not because it delivers anything much, but because it brings into sharp contrast conventional ideas about consciousness (based on materialism), and the ideas that interest us. Originally I was fascinated by the concept of AI, and even had some involvement, but I could see how empty it was before its crash in about 1990 (when people suddenly got interested in Neural Nets, which seem to deliver in certain areas, such as face recognition - which is something we don't seem to do consciously).

That article seemed to imply that the idea of self driving cars was being abandoned (discussed in the comments, which are worth reading). I have argued several times here that self driving cars can only work in artificial conditions, or with impossible restrictions of one sort or another. If the huge investment in this technology does get nowhere it will be very telling - particularly because the materialistic concept of AI keeps grabbing people's imagination, but never really delivers.

David
 
If we look at the various points that are raised there, most seem to relate to performance, but I would argue that performance isn't the primary issue - there is simply no way to solve the problem however many computer resources are thrown at it. Let's deal with the numbered points one by one:

Difference # 1: Brains are analogue; computers are digital

IMHO, this is a non-issue. Computers may be digital, but they handle floating point numbers just as well - in fact far better - than any analog hardware - which is why analog computation has fallen out of fashion.

Difference # 2: The brain uses content-addressable memory

This can only be a conjecture at this time, and CAM can only provide a speed up - a suitably fast conventional computer can simply search through the memory for the relevant data. Software developers use all sorts of tricks to improve the performance in typical cases.

Difference # 3: The brain is a massively parallel machine; computers are modular and serial

Once again, this is only a performance issue. In general, parallel computation is much harder to specify that sequential operation.

Difference # 4: Processing speed is not fixed in the brain; there is no system clock

Why would that be a gain - designing a computer that way would be a nightmare!

Difference # 5 – Short-term memory is not like RAM

This is not really an issue at all. It only seems real because computer people use the word 'memory' to refer to something that is rather low level. However software can build all sorts of complex entities over the top of this.

Difference # 6: No hardware/software distinction can be made with respect to the brain or mind

This seems a completely useless distinction, except that the article clams that the mind emerges from the brain! IMHO this claim is too vague to consider further.

Difference # 7: Synapses are far more complex than electrical logic gates

Again this is irrelevant. Logic gates are the lowest level of a computer, and all sorts of higher level constructs are built on top of them.

Difference #8: Unlike computers, processing and memory are performed by the same components in the brain

Again, this doesn't offer any obvious advantage.

Difference # 9: The brain is a self-organizing system

Programs can do operations that are equivalent to self organising. For example, there are neural net models that can do stuff like this, but in practice they are usually implemented on a computer for convenience.

Difference # 10: Brains have bodies

This is a sign of desperation - think what Steven Hawking's mind can do, despite his body being more or less defunct. Besides, if a body were required, a computer program could surely simulate one. Perhaps this is really a sneaky way of saying that there is something special about biological matter - something that would upset conventional biologists :)

Bonus Difference: The brain is much, much bigger than any [current] computer

Originally AI used the excuse that it had to wait for sufficient computer resources. This excuse more or less dried up in the 1990's, because by then there was more resource available than people knew what to do with!

David
 
Last edited:
Why isn't a program that can perform better than a human (like a chess program, a Jeopardy program, or a medical diagnosis program) AI?

I'd hazard a guess it's because tasks like this can be represented by algorithms. Executing an algorithm, however slowly, is only a portion of what true intelligence can do. Anyone who's done any programming knows that computers possess no discrimination. They simply obey the syntax of a programming language even when they generate very quickly, from the programmer's viewpoint, results meaningful to him -- or garbage for that matter.

Nearly forty years ago, they were described to me as TOMs or Totally Obedient Morons. They'll do precisely and only (barring hardware or firmware glitches) what they've been programmed to do, no more and no less. There's no way they can do anything novel: everything they do is totally governed by the programmer via his program.

Computers can only ever execute the intention of a programmer insofar as that can be represented by a programming language. They have neither intuition nor imagination, two vital aspects of intelligence. To even simulate these, one has to find a way of stating them in algorithmic terms. As soon as one does that, true imagination and intuition become impossible.
 
I'd hazard a guess it's because tasks like this can be represented by algorithms. Executing an algorithm, however slowly, is only a portion of what true intelligence can do. Anyone who's done any programming knows that computers possess no discrimination. They simply obey the syntax of a programming language even when they generate very quickly, from the programmer's viewpoint, results meaningful to him -- or garbage for that matter.

Nearly forty years ago, they were described to me as TOMs or Totally Obedient Morons. They'll do precisely and only (barring hardware or firmware glitches) what they've been programmed to do, no more and no less. There's no way they can do anything novel: everything they do is totally governed by the programmer via his program.

Computers can only ever execute the intention of a programmer insofar as that can be represented by a programming language. They have neither intuition nor imagination, two vital aspects of intelligence. To even simulate these, one has to find a way of stating them in algorithmic terms. As soon as one does that, true imagination and intuition become impossible.
Has anyone claimed to have produced a computer with intuition or imagination?
 
Has anyone claimed to have produced a computer with intuition or imagination?
The point is, they imply that AI is somehow different from a useful program. In one sense it is a best thought of as a marketing trick, but some academics actually believe in the trick - i.e. that you can create a program that actually can behave like a human is some sufficiently broad domain (that bit is crucial).
Why isn't a program that can perform better than a human (like a chess program, a Jeopardy program, or a medical diagnosis program) AI?
Well for the concept of AI to have any real meaning, it has to consist of some collection of techniques that distinguish it from ordinary programs - which routinely exceed the performance of humans in specific domains. There is no heart of a chess program that you could extract and put to use elsewhere - solving crossword puzzles, for example

David
 
The point is, they imply that AI is somehow different from a useful program.
Who? Where?

Well for the concept of AI to have any real meaning, it has to consist of some collection of techniques that distinguish it from ordinary programs - which routinely exceed the performance of humans in specific domains.
Like a completely unreliable, malleable memory system? ;)
 
https://www.theguardian.com/technol...-intelligence-ai-fukoku-mutual-life-insurance

Japanese company replaces office workers with artificial intelligence

Insurance firm Fukoku Mutual Life Insurance is making 34 employees redundant and replacing them with IBM’s Watson Explorer AI
...

The system is based on IBM’s Watson Explorer, which, according to the tech firm, possesses “cognitive technology that can think like a human”, enabling it to “analyse and interpret all of your data, including unstructured text, images, audio and video”.​
 
When I was an office worker I used coloured pencils to maintain employee holiday records. The guy sitting next to me used a mechanical adding machine with a handle to pull for each addition. Seriously, I'm not joking.
Jeepers. How old are you?
 
Jeepers. How old are you?
'Bout your age. I just come from a backward part of the world ;)
Well at the time I was working for a major multi-national chemical company. But there were still backwaters that time forgot.

Forgot to add - I had a bicycle to deliver internal company mail - we'd done away with the ponies by the time I started there.
 
Last edited:
I'd almost forgotten - that was a strange place to work. Occasionally I'd find myself in the office completely alone. And then I'd start to hear background sounds. There was a sort of hissing sound which sounded eerily like someone breathing. There were pipes for hot water radiators but that didn't seem to be the source. I thought it might have been the fridge we used to keep milk, but whenever I went near the fridge I heard nothing at all. I'm willing to accept that it was some industrial process making the breathing sound, but I could only hear it inside that building when alone, nowhere else on site. When I sat still it just seemed to get louder, but there was no sense of direction.
 
'Bout your age. I just come from a backward part of the world ;)
Well at the time I was working for a major multi-national chemical company. But there were still backwaters that time forgot.

Forgot to add - I had a bicycle to deliver internal company mail - we'd done away with the ponies by the time I started there.

So... No jousting then?
 
https://www.theguardian.com/technol...-intelligence-ai-fukoku-mutual-life-insurance

Japanese company replaces office workers with artificial intelligence

Insurance firm Fukoku Mutual Life Insurance is making 34 employees redundant and replacing them with IBM’s Watson Explorer AI
...

The system is based on IBM’s Watson Explorer, which, according to the tech firm, possesses “cognitive technology that can think like a human”, enabling it to “analyse and interpret all of your data, including unstructured text, images, audio and video”.​
That is simply replacing people by some software! The label AI is entirely arbitrary.
The technology will be able to read tens of thousands of medical certificates and factor in the length of hospital stays, medical histories and any surgical procedures before calculating payouts, according to the Mainichi Shimbun.

While the use of AI will drastically reduce the time needed to calculate Fukoku Mutual’s payouts – which reportedly totalled 132,000 during the current financial year – the sums will not be paid until they have been approved by a member of staff, the newspaper said.
That doesn't sound terribly hi-tech, and notice that a human is used to check the result before a payout is made!

David
 
That is simply replacing people by some software! The label AI is entirely arbitrary.

That doesn't sound terribly hi-tech, and notice that a human is used to check the result before a payout is made!

David
It actually sounds like a very conventional computer system.

The difference perhaps is the input stages are accepting a wider variety of media, rather than using a rigidly-defined input format.
 
Have any of the coding guys on here any idea how you'd start to recreate non-digital human memory with all its shortcomings?
 
Hi, Malf

Thanks for the kind replies to my questions. Very enlightening !!
I can't remember any questions... I may have missed them in all the rhetoric and pontification so please refresh me. ;)
 
Back
Top