Artificial Intelligence: You know it isn't real, yeah?

"Where's the intelligence?" cried a voice from the back. It's not quite the question one expected during the Q&A session at the end of the 2019 BCS Turing Talk on Artificial Intelligence. The event was held earlier this week at the swanky IET building in London’s Savoy Place and the audience comprised academics, developers and …

[John Geek]

I've always called AI "Artificial Ignorance" and I've seen nothing to date to persuade me otherwise.

[I.Geller]

IBM Watson? Waymo? Google and Yandex translation?

AI structures texts in patterns, where all patterns of the texts specify the context and subtext.

[H in The Hague]

"Google and Yandex translation?"

Google Translate (GT) is actually a good example of where pattern matching fails. Its apparent translation is created by matching patterns in the source language with those in previous translations into the target language - a purely statistical process, devoid of any understanding, any intelligence.

That often leads to translations which read surprisingly well, but may be incorrect. Example: a client provided me with a document in Dutch and mentioned they'd had that translated from French. It seemed to be perfectly good Dutch, but one sentence would have made much more sense had it included the word "not". That made me suspect they'd used GT or similar. So I requested the original French doc and even with my extremely rusty secondary school French I could see that the original document did indeed include 'not'.

Tested it: GT French -> Dutch left 'not' out while GT French -> English correctly included 'not'.

Another problem with GT is that it feeds off source texts and translations it finds on the Web - but increasingly those are its own work, hence you get the snake biting its own tail, reinforcing its mistakes. GT can be a useful tool but one to be used with caution by those who are aware of its workings and shortcomings.

[I.Geller]

Agree. The problem is that they don't have personal profiles, which contain used by the person texts and their structured representations. That is, all available texts are used instead of those used by a particular person. But still there is some result.

[Herby]

Isn't!

Well that is what "Artificial Intelligence" is these days.

Pretty simple if you ask me.

[Muscleguy]

Recidivism Fail

I expect there are a legion of criminologist out there who spout 'recidivism' all day and every day without pause. What a stupid comment from someone who sneers at how English as a second (third? fourth?) language speakers pronounce 'algorithm'. Which like al cohol is an Arabic word originally and the al is a separable prefix, so al-gor-ithm is perfectly correct for many.

That's just the ignorances so far, let's see how many more I can find . . .

Sometime British scribes might learn that it is not necessary to sneer at others when writing articles.

[Toni the terrible]

Re: Recidivism Fail

It's not sneering, its their cognative bias

[Captain Kephart]

AI is an approcah, not an outcome

The scary thing is that many politicians and opinion-formers really think that current machines are 'intelligent' enough for humanity to let them make decisions for us ... and the machines have no notion of that.

The best definition of intelligence I ever heard was from Prof Igor Aleksander (who had a face-recognition and speaking neural network at Brunel University in the UK in 1983). He said there is no such thing as 'artificial intelligence' - just intelligence. He felt that the problem with AI had / has been that its practitioners thought that it was something you programmed – of the style of:

FOR 1 to n; BE INTELLIGENT; LOOP

and that this was always nonsense.

The Six Laws of Intelligence

Instead, Igor said (I am paraphrasing his deep discourse), you have intelligence when you:

1) are self-aware, and aware that you are self-aware;

2) able to sense the world and other beings and perceive that they are self-aware;

3) can appreciate that they have different motivations and views of the world to yourself;

4) can conceive of what their view(s) of the world may be;

5) can reason from those points of view and synthesise them with your own ...

6) and lastly be able to act, interact, and effect change in the world in line with those things - anticipating, adapting and changing over time - and so changing the nature of your intelligence in line with the real-world context.

The various kinds of simulations and emulations of ‘intelligent’ behaviour succeed as far as they do because of the human ability to anthropomorphise and attribute intelligence where it does not exist (think of Tamagotchi as a more extreme example). We even do it with objects in our homes (such as cuddly toys).

There is only Intelligence - AI is an approach not an outcome. This is because intelligence is really a social phenomenon (not an individual property) arising out of meaningful and reciprocal relationships over time – even the famous ‘Turing Test’ is set in a social context - and computers have no idea about that, and are nowhere near achieving it.

Don't get me started on Alexa, Siri etc ... "Alexa, review and edit this post for me." ...

[I.Geller]

Re: AI is an approcah, not an outcome

What you see today and call AI is only its basis, which was originally intended for commercial use - how to find information. That is, it was based on the definition of NIST TREC definition: AI must answer both Factoid and Definition questions.

Google appeared on the earlier version of AI, which used the traditional n-gram parsing. The difference between this AI from which Google uses - in the new AI-parsing.

The new AI-parsing allows AI to structure texts into patterns, each of which is a direct analog of a programming language command, and which are contextually-subtext targeted. Then Machine Learning is the addition of new texts, if the old ones do not have the necessary patterns.

If you want AI to be "able to act, interact, and effect change in the world in line with those things" - AI can.

Now on Intelligence. It's into tuples, where in mathematics, a tuple is a finite ordered list of elements. In other words each pattern must be annotated/ explained because otherwise it cannot be found/ is not unique. Therefore our brain is biological computer which keeps sets of tuples, and AI does the same. Yes, there is no such thing as 'artificial intelligence' - just intelligence.

[Toni the terrible]

Re: AI is an approcah, not an outcome

Therefore small children are not intelligent

[I.Geller]

Re: AI is an approcah, not an outcome

Intelligence is a process. If you use External Relation theory's postulates and consider the child as a constant - no, he is not intelligent. For Internal theory he becomes older and intelligent.

[I.Geller]

Re: AI is an approcah, not an outcome

All the chatterers, who commented on this article, are trying to appeal arithmetic to differential functions. That is, intelligence is about becoming - and any attempt to stop and define it (in the term of constants) doomed to disaster.

AI was created as a differential function which becomes, it cannot be described in the finite terms:

- AI is sets of tuples, where each tuple is a number of phrases;

- while, the most vivid example, SQL operate with words and never with the tuples as sets of phrases.

In other words, AI is a continuously changing set of tuples, while all without one exception, the existing theories operate with separate words, constant.

[Michael Wojcik]

Re: AI is an approcah, not an outcome

even the famous ‘Turing Test’ is set in a social context - and computers have no idea about that, and are nowhere near achieving it

There are chatbots which have beaten human judges in Imitation Game (aka "Turing Test") challenges. Those challenges are inevitably limited - they have time limits, at least - and given, say, several months to interact with members of an Imitation Game panel the judges would probably eventually distinguish the participants correctly, at least with decent probability. But under the terms in which those contests were conducted, the 'bots won.

People who actually work in AI / ML are not particularly interested in those results, because they're not particularly interesting. Turing didn't intend for people to hold real Imitation Game events. It's a philosophical thought experiment.

Basically, it's an argument for the sort of view of intelligence that might emerge from the American pragmatist school of epistemology: we know an entity X is a member of class Y because it exhibits the visible attributes of members of that class. We treat things as black boxes and concern ourselves with how they interact with the world.

It's interesting to contrast Turing's position with John Searle's in his Chinese Room argument, which is essentially a logical-positivist and phenomenological one. Searle says, in effect, "I'm not sure exactly what I mean by 'thinking', but this description of what one approach to AI is doing isn't it". (Logical positivism asks "what do we mean by the term 'X'?", and phenomenology asks "what are we doing in our minds when we do Y?".) So Searle does want us to consider what's happening in the box, and whether we think it might be similar to what seems to happen in our minds.

It's mildly ironic that the Englishman Turing leaned toward an American philosophical school, while the American Searle toward one most closely associated with the UK. But then we hope our better thinkers will reach outside of whatever's popular in their own playgrounds.

Robert French, among others, has pointed out (in a piece in CACM some years ago) why the Imitation Game isn't a useful practical test of machine intelligence. Appealing to it at this point in the game doesn't really help, except as a touchstone to illuminate your philosophical position.

Aleksander's definition of intelligence which you summarized above won't satisfy everyone, but it's one that can be argued for. I don't have any great objection to it myself, though I'm not ready to endorse it either. It's interesting to note that it combines logical-positivist and phenomenological criteria (items 1-5, for the most part) with a pragmatic one (6).

[I.Geller]

Re: AI is an approcah, not an outcome

There must be a clear separation between thinking and its result.

Thinking is a process, a function; and its result is the limit of the function - the moment when the function loses its character and becomes something completely different.

AI, for example, understands the paragraphs in the sense that they contain complete thoughts, that is, they are integrals in their relation to thinking; where dictionary definitions for the paragraphs' words form constant when integrated.

[Charles 9]

Re: AI is an approcah, not an outcome

"AI, for example, understands the paragraphs in the sense that they contain complete thoughts, that is, they are integrals in their relation to thinking; where dictionary definitions for the paragraphs' words form constant when integrated."

And therein the human condition can throw it off. What if it's a poorly-written paragraph that never comes to a concise point?

[I.Geller]

Re: AI is an approcah, not an outcome

The quality is very important, a paragraph can be equated to a dictionary definition for all its words.

AI technology first of all is a search for information technology, it appeared as the answer to NIST TREC QA challenge. For example Google searches using the earliest version of AI, which had not AI-parsing technology and couldn't convert paragraphs into dictionary definitions. Thus Google cannot search by sense (that is to structure the paragraphs), and completely rely on popularity.

The quality is essential! The better the paragraph the better it defines all its words' meanings/ create right tuples (where a tuple is a finite ordered list of the paragraph's phrases)

[Ted's Toy]

I would like to have real intelligence

All this talk of AI when I went to school in the dim and distance past we were taught that the real thing was what one needed, not an artificial something.

[amanfromMars 1]

Re: I would like to have real intelligence

As a worthy substitute and wonderful surprise, is greater imagination a blessing to be constantly thoroughly worshipped in Effective Tempestuous Admiration of the Mighty Steal.... Absolutely Fabulous Imagination Leading what's Real and Realisable in Proprietary Intellectual Property Hosting Communities.

Everything Good with Nothing Bad would Immediately Server a Perfect New Beginning in a Wholly Novel Different Direction.

Something easily hosted and programmed for Present Media to Squawk is Everywhere is their Trafalgar Turing Test.

All this talk of AI when I went to school in the dim and distance past we were taught that the real thing was what one needed, not an artificial something. .... Ted's Toy

Absolutely Fabulous Imaginations are Raw Core Virgin Source Ore for Refining and DeepMetaData BaseMining. Beware the dragons and goons and say Hi to All Friendly Looney Toons is good sound advice to follow to steer well clear of so some really soggy boggy territory with a persistent grip/unholy attraction/devilish interest/insatiable reward. They be Heavenly Dues, Surely.

What/Who Leads your Current, Intelligence Communities? Are there UNSeen Hands, Stout Hearts and Greater Minds at Play for Y'All from Today and Henceforth?

:-) And does that Generate Permanence for a Persistent Presence? And Virtual AIRealisation of All of the Aforementioned Facts for Future Fictions delivering Naked Virgin Territories for Popular Colonisations ....... Exciting Future Builds. .... you know, that Deep See and Proud Vision Stuff of Kings and Queens where Perfect Adams Adore Serving and Servering Immaculate Eves Desires and Wishes to Mutually Satisfying EMPowering Satisfaction.

Have the Key to that Store and Heaven knows it can wait whilst all Hell breaks loose as Future Derivatives and Virtual Options are considered a Must Have Vital Commodity.

[[e-mail]]

Artificial, as in Fake Intelligence

What people have been told to expect from Artificial Intelligence is that a computer will become self aware, and become artificially like a human being.

But that is not the case, nor ever can be. That is because a computer has no instincts, emotions, autonomic nervous system, or anything remotely alive or possible to become sentient, ever. All artificial intelligence really is, is what ever a programmer decides to put into his fake simulation. And that can never be real sentience because no programmer likely will ever know how to do that.

Let me give you an example. If you say the word "dog" to an English speaking human, they will receive all sorts of associated memories, data, images, etc., that will include the dogs you have seen, read about, etc. But if the word "canine" is used instead, you likely get the same associated responses. In fact, you can use another language even, and it won't matter. That means this is not at all like a database program based on text keys. Guess what? That means humans are BORN with a built in semantic. We all internally have some unambiguous internal representation for dogs, that is identical in all humans.

So if you understood that, then you would understand that before we could actually ever duplicate what humans do naturally, we would have to somehow figure out how humans do it. And that likely is never going to happen. So forget about artificial intelligence. It is not likely ever going to happen.

[Charles 9]

Re: Artificial, as in Fake Intelligence

"Guess what? That means humans are BORN with a built in semantic. We all internally have some unambiguous internal representation for dogs, that is identical in all humans."

I disagree. We only perceive this because we normally associate with people who are much like us: seen much of what we've seen, including dogs. But what if you head out to the sticks, to peoples who have such limited experiences that they may not recognize such simple things as a pet dog or a housecat...or even a ball. About the only things we recognize solely on instinct (tested on newborns who have the least life experience possible) is another human (and that's likely a survival trait).

[Michael Wojcik]

Re: Artificial, as in Fake Intelligence

That is because a computer has no instincts, emotions, autonomic nervous system, or anything remotely alive or possible to become sentient, ever

OK, now explain how human beings have instincts, emotions, and sentience.¹ Are our minds not effects of mechanical processes? What prevents us from creating an artificial device which produces similar effects?

(And I have no idea why you lumped "autonomic nervous system" in there.)

¹Perhaps you mean sapience? It's easy to argue that some machines already have sentience.

[[e-mail]]

No AI, But People Expect Autonomous Cars?

All the post here seem to pretty much agree that there is no such thing as Artificial Intelligence, and it will be too difficult to ever come close to what humans do so easily. But then what is so strange is that for some reason people have such a desire for autonomous vehicles that they think it is actually possible or even happening right now. I assume you that there is no such thing as autonomous vehicles. They are all fakes running on GPS, and can not recognize or read street name signs, know where lanes, are or recognize turn signals or brake lights. So then why is it people have this unrealistic disconnect? Perhaps we are lazy or incredibly gullible?

[Easy E]

Re: No AI, But People Expect Autonomous Cars?

The other aspect people fail to consider is that these cars currently operate in only good weather. They aren't subject to (un)intentional sensor attacks. The systems are typically tied into the infotainment system (seems like that would be a clearly obvious no-no for a plethora of security reasons). There isn't a public prediction tree for which decision a vehicle will make if it comes across a bad situation, if the choices amount to running into an oncoming car, ditch or tree when it detects an 'accident' situation. What bad choice will the vehicle make and why? Will a semi hit a bus head-on as opposed to striking a bridge column? Will it strike a pedestrian or hit a concrete power column? Will it drive through a wild fire because it doesn't know any better? Will it stop and turn off if exhaust gets into the car's cabin? How will it know to drive at a slower rate of speed if there's a risk of black ice? What is an acceptable rate of failure, because auto manufacturer's aren't going to be held to 99.999% as it relates to faults?

[Michael Wojcik]

Re: No AI, But People Expect Autonomous Cars?

All the post here seem to pretty much agree that there is no such thing as Artificial Intelligence

"Several people who seem to agree with me seem to agree with me. Must be true!"

[Anonymous Coward]

Real world manufacturing and A.I.

Working in a manufacturing sector with metal machining as the primary function at the facility I am currently located, it becomes annoying hearing stock pundits state (with absolute certainty) that these manufacturing jobs will be replaced in a matter of years with A.I. As if... we have problems with simple automation due to tight tolerances. Human monitoring and interaction are required even if the machines are running properly because the machines can't predict how different parts will match up because there are too many variables. Only a few of us can consistently recognize how each of the machines will behave and it's difficult enough to get a skilled person to understand more than the production line they're assigned to.

[Roy Lofquist]

Douglas Adams

Douglas Adams had thoughts about AI. This, from "Dirk Gently's Holistic Detective Agency", gets at the difference between mere deduction and mysterious intuition:

“Sir Isaac Newton, renowned inventor of the milled-edge coin and the catflap!"

"The what?" said Richard.

"The catflap! A device of the utmost cunning, perspicuity and invention. It is a door within a door, you see, a ..."

"Yes," said Richard, "there was also the small matter of gravity."

"Gravity," said Dirk with a slightly dismissed shrug, "yes, there was that as well, I suppose. Though that, of course, was merely a discovery. It was there to be discovered." ... "You see?" he said dropping his cigarette butt, "They even keep it on at weekends. Someone was bound to notice sooner or later. But the catflap ... ah, there is a very different matter. Invention, pure creative invention. It is a door within a door, you see.”

[loco_wunee]

Define intelligence? Ok...

Here's how I will know that a robot is as smart as me: My ideal robot would, without flinching, without prompting, immediately scratch any itch on my body before my hand can reach. It will scratch vigorously, sufficiently, and comfortably any region of my body, on or under the skin, and with such discreetness and moral sensitivity that I can take this robot to any social gathering, I can be in any state of dress, and never slow me down.

Show me the scratch.

Then we'll talk about intelligence.

[amanfromMars 1]

Re: Define intelligence? Ok... how about Knock Out Future Opportunities

..... in Virtual Space Control Space

How about, loco_wunee, Virtual AIMachinery whose Ardent Attention to Satisfying Desires with Insatiable Needs and Prime Feeds is Practically Almighty.

And Ideal for Pleasure Robots Servering to Venus her Bounty and Immaculate Captures/Free Will Slave Followings to AIMetaPhysical Unions ..... Cleared Cyber Spaces.

In those places, loco_wunee, you do what you want to unfold what you need to feed and seed and surely concede is a Secure AI Future Supply, and for Secessions*, a Perfect IntelAIgent Partner to be in Bed and Embedded with.

* Traditionary Hierarchical Structure Implosions

And that's only scratching at what's further there, l_w. Such would normally constitute an Earthly Pow Wow .... some United Nations ProAction ...... advising of Ongoing Strange Virtual ACTivities in which they Require Immediate Immense Assistance.

The reply as to whether they do or they don't is irrelevant, for both are adequately answered with a perfectly honourable replies to the Posit.

Now, tell me that is not a disturbance in the force, Mr Beale?:-) ..... https://youtu.be/yuBe93FMiJc

[Toni the terrible]

Re: Define intelligence? Ok... how about Knock Out Future Opportunities

You are surely a Man from Mars

[CreActive]

Broken link

Moderators: the COMPAS link points to the wrong article

[I.Geller]

Again

Intelligence is based on sets of tuples; where in mathematics a tuple is a final ordered list of elements (patterns, phrases). The longer the tuple - the more unique it is, it can be more easily distinguished and found. Thus, the brain has the tuples and the connections between them, and intelligence is the ability to find the right tuples and apply them. If there are none - a search and addition for new tuples performed.

This is our intelligence and AI basics.

Stop to fantasize and begin to read what is already published! You all look ridiculous!

[John Brown (no body)]

I'll believe...

I'll believe we are at the beginnings of AI when mapping software is capable of telling me that the nearest $Store location is the one within 3 miles by road of my location instead of the one half a mile away on the other side of the river that requires a detour 3 miles down river to the tunnel then 3 miles back up river, or the alternate route 5 mile up river to the bridge then 5 miles back down river.

[Glenturret Single Malt]

Recidivism my algorithm

The "cidivism" part of recidivism is pronounced with the same emphases as algorithm (and its anagram logarithm) so if you can pronounce one you pronounce the other. (OK, I know it was a joke, but just saying).

[itzman]

The fundamental problem is that...

...Bias and prejudice are efficient.

See a snake? Kill it or run like heck. Who CARES that 80% of snakes are perfectly safe. Killing a safe one doesn't harm you. Cuddling a rattler does. Regarding all snakes - or indeed mushrooms - as poisonous saves you having to carry around a catalogue of the very few that are not.

Most [issues] are caused by [a few identifiable members of set x] is most easily encapsulated as

All members of [set x ] cause [issues]

What we are seeing in this is the prime example of 'its not fair' versus 'it doesn't work'

E.g. if you want to halt the spread of Islamic fundamentalism and radicalisation, ban the religion, beards. burkahs, niqabs, hajibs , imprison anyone who preaches it and shut down any mosque or website that carries any materials. etc.

Unfair, but effective,..

The human mind seeks to use pattern recognition to arrange the world into 'objects' that have 'generic properties' . So it can apply generically effective general rules without having to examine the particular.

There are those who are stupid enough to feel ashamed of their propensity to do this and project the negative aspects of this onto others.

Don't be one of them.

Wisdom comes from accepting and then making allowances for the fact that we are all prejudiced and biased, and if we were not we would have eaten the poisonous mushroom years ago.

Those who claim it is others who are bigoted, are usually the worst bigots themselves.

One thinks instantly of the jackboot mentality of self styled 'anti-fascist' organisations.

[Anonymous Coward]

Re: The fundamental problem is that...

Until you realize those "efficiencies" come at a price that you may not realize at first.

Kill that "safe" snake? Now you're gonna have a rodent problem because you knocked off one of their more-efficient predators.

Prejudice against a whole bunch because of a few bad apples? You've just tarred everyone with the same brush. Be prepared to deal with more of the same. If people say they're evil, they may as well become evil.

[DerekCurrie]

It's advanced Expert System algorithms

Thank you Alistair for hitting the important points about what is euphorically, unrealistically called 'Artificial Intelligence'. "AI" is just another meme being used by marketing divisions and self-promoting researchers as bait to sell what are actually advanced Expert System algorithms.

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

Expert system design was introduced in 1965. It has been slowly advancing thereafter. What has triggered the 'AI' moniker are actually advances in speech recognition (speech, to text, to bits) and the reverse (bits, to text, to speech). Much of the speech recognition work was accomplished by Dragon Systems, starting in 1975. Dragon is currently owned by Nuance. As far as I am aware, all of the current popular 'AI' systems involving speech recognition license base technology from Nuance. Apple and IBM, among others, developed their own speech recognition systems in the 1990s. But they fell by the wayside. It is said the Google and Microsoft have been developing their own speech recognition systems. However, I strongly suspect they both use Dragon technology.

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

The general concept of Expert Systems is to take a provided input and efficiently traverse an available database for the best matching output. What has become interesting in recent years has been what's called 'machine learning' whereby the algorithms are able to collect new data and add it to their source database over time in pursuit of providing better and more up-to-date output as answers to input queries.

In other words, there is no 'intelligence' at all apart from that imbued by the euphoric, the meme entranced, marketing executives and researchers in need of academic recognition. We may in fact never be able to create anything that is actually 'intelligent' in the science fiction sense. The first goal of artificial intelligence is to be able to interact with it and not be able to tell if it is human or machine. This is called the Turing test. Thus the connection of AI to Alan Turing. And no, so far there has been no so-called 'artificial intelligence' system capable of passing the Turing test. There have been rumors of such a system for decades. But none of them have qualified. No doubt, with time the term 'intelligence' will be bent and twisted to fit the latest attempt at AI. However, the technology we currently have better qualifies as Artificial Idiocy with the goal of making it as mildly idiotic as possible.

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

ETHICS: As ever, technology is a tool for the betterment of mankind. Any technology used instead as a weapon of murder and destruction is no longer a tool but an abomination, a victim of self-destructive, territorial and tribal instincts still persistent in species Homo sapiens sapiens ('wise wise', as we wish we were). I call it Coward Murder Machinery, as currently exemplified by killer military drones controlled by humans from a distance with no danger to themselves. Another step in coward murder machinery will be autonomous drones and other robots whereby their creators and controllers will attempt to dodge responsibility for their outcome. Don't be fooled. Mankind will always be responsible for whatever technology we create, including all its consequences. We will become actual 'sapiens sapiens' when we achieve the wisdom to stop murdering one another for any reason.

Conclusion: Ethics begin and end with the creators, programmers and users of technology, not the technology itself. IOW: Don't blame Hal 9000. Blame its inventors and programmers. Hal 9000 or any other 'AI' is just a tool.

https://en.wikipedia.org/wiki/HAL_9000