This is a good balanced article that gets a lot of things right. We should take a forgiving approach when we talk about AI systems. And as the author points out the problem is not that AI systems dont have understanding yet. The problem is with the hype which leads many to believe that we are close to building systems which can understand us.
That said, I have a small problem with the examples presented to say that already machines understand us :)
The article says 'For example, when I tell Siri “Call Carol” and it dials the correct number, you will have a hard time convincing me that Siri did not understand my request"
Let me try to take a shot at trying to explain that Siri did not "understand" your request.
Siri was waiting for a command and executed the best command that matched. Which is, make a phone call.
It did not understand what you meant because it did not take the whole environment into consideration. What if Carol was just in the other room. A human would maybe just shout "hey Carol, Thomas is asking you to come", instead of making a phone call.
If listening to a request and executing a command is understanding, then computers have been understanding us for a long time. Even without the latest advances in AI.
> It did not understand what you meant because it did not take the whole environment into consideration.
This is the crux of the matter. These voice recognition agents are trained with goal of accurately modelling a function that converts recorded sound to a series of words, and then act on those words to perform the most appropriate action. They are NOT trained to model the entire world, which is an incredibly complex task that no one has been able to formulate as a problem that computers can solve, yet. Humans on the other hand, have a machine that is extremely well-equipped to do just that - the brain. And that is exactly why humans are able to "understand" things, while we feel that machines are not, with our definition of "understand".
In the far distant future, if and when we do figure out a way to model the entire world, come up with suitable objective function, and solve it on a computer, there's no reason why that machine should be any less capable of understanding things than the average human.
I think this is partly down to us humans defining "intelligence" as "like us".
We have a very specific set of evolved traits that define our understanding of the universe. A lot of that is social. So our "understanding" of the phrase "call Carol" includes a wide range of social cues about what that means, and your example is perfect: "call Carol" means that I want to talk to her, and that would be better done in person if possible, but that "if possible" has a more-or-less specific range of "if she's within earshot so I can yell for her", which is limited to the range of a human voice (but not the maximum range, like screaming, but just a normal yelling range). Which is less if the door is closed, or there's music playing, or Kevin is trying to nap in the other room. And not at all if we're in a library, or concert, or even a public space where yelling would draw attention. If "call Carol" has to include all of these to qualify for "understanding" then I think I know some people who fail at this test.
My go-to thought experiment on this is Dolphins. Dolphins are intelligent, have language, etc. But their understanding of the world must be so different. Trying to explain to a dolphin what "tripping someone up" means is going to be tricky. They may understand the words, but they'll never understand the concept.
We swim in a sea of social cues and non-verbal communication. We can program an AI to imitate more and more of this, and be aware of more of it, but it's like teaching dolphins about long-distance running. It's never going to come naturally. And they're never going to evolve that understanding naturally (like we do as children) because it's not in their nature. We anthropomophise our machines a lot, and we assume that they'll grow (like children) to grok all of our social cues eventually, because our only experience of similar situations is, well, children. But they're just machines, designed for a single purpose. They're never going to grok this. They're never going to be "like us" and really understand all the social ramifications of "call Carol". At some point I think we're going to have to accept this, and say that the machine understands the phrase "call Carol" enough. TFA draws the line at the machine calling Carol, and that seems reasonable.
So the next version of Siri can locate Carol's phone in the next room and will just beep her phone to tell her to see you. Of course that's still not understanding.
Which is an absolutely textbook example of begging the question.
If you could make a machine pass the Turing test it might be intelligent - but no one has, and it's debatable if it's even possible, and it's even more debatable if, hype notwithstanding, the Turing test is even a good test of human-equivalent intelligence, because it ignores side channels that are fundamental to human communication, including tone of voice, posture, and facial expression.
(Yes, people communicate over email/SMS. But no one communicates over email/SMS without an implied social context that hugely limits and simplifies the content of any conversation.)
It's not the "call Carol" problem that needs to be solved. It's the "understand the entire world context well enough to know how to call Carol without being told - which includes being able to research information that isn't already available, and also includes edge cases like 'We went to Carol's funeral last week' and 'Carol had her phone stolen yesterday' and 'Carol is flying to Australia and won't be receiving messages for another 12 hours" and "Carol prefers FaceTime to WhatsApp."
And so on.
Ultimately your toy machine has to show evidence that it understands the entire world and can learn about it like a human can - which includes being able to do original research that isn't a simple literal Google search, parse humour, understand emotional responses and common cultural references, and follow standard social protocols.
That's a much harder problem than having a vaguely plausible limited text-only conversation, whether it's in Chinese, English, or Swahili.
I would call the moving goalposts a subtle sign of a win as well if it is getting closer that previous "unthinkable" tasks need more qualifiers. Missing or adding them makes it easier or harder. To be a smartass anything can pass a text messages from the comatose (that is nothing) and nothing can reliably "prove" itself god by say resurrecting and teleporting your dead relatives to you would be obviously useful but impossible as that isn't something text messages can do.
> Siri was waiting for a command and executed the best command that matched. Which is, make a phone call.
ISTM there's no more "understanding" involved in this than when I touch the Contacts icon on my screen, then "C", "A", "R", etc until Carol's entry is displayed, and then I touch the Phone icon to initiate a call.
The fact that the interface used was sound-waves that the device recognised as matching the keyword "call" and the contact-list entry "Carol", rather than my finger touching specific areas of the screen, may be a handy feature. Of course it's a triumph of signal processing, fuzzy recognition, etc. But there's no more "understanding" involved than in the touch-screen version of the action, or in typing a command and parameter into a terminal window.
> If listening to a request and executing a command is understanding, then computers have been understanding us for a long time. Even without the latest advances in AI.
I think this is a reasonable thing to say, in the limited way he has defined ‘understanding’. People forget what a titanic achievement that user interfaces that allow us to communicate our intentions to a computer and receive a relevant response actually are, whether it’s using a voice or clicking a button.
Worse, we've build a conceptualisation of these phenomena that renders us as calculators on steroids! I don't think it's feasible to account for or describe anything more with the ideas we have so far.
I have a straightforward definition of "understand". To understand means to be able to give a (representative) example of the (intensionally) given set. Though it is harder than it seems, as it usually means solving the constraint satisfaction problem.
For example, take the classical AI knowledgebase fragment, "bird is animal that flies". If I ask example of bird, it can say "eagle", and exhibit some understanding. We can then probe further and ask for a bird which is not an eagle. If it says "bat" or "balloon", it exhibits that it still doesn't understand birds quite right.
In particular, if the description is nonsensical and thus impossible to understand, we cannot give any examples.
This idea was really inspired by the study, where they asked people to recognize nonsensical and profound sentences, describing certain situation. The profound are the ones where you can create a concrete instance of the situation.
You've rigged this up to operationalize it for current digital machines.
"Understanding", "Intelligence", etc. is a feature of animals in their environment. We need to begin there; and that is what we are talking about.
We "understand" how to drive as a dog "understands" how to play fetch. Understanding is not ever going to be a trivial rule that some digital system may instantiate.
It will always require direct causal contact with an environment. In my view "understanding" is "competent play in a changing environment" -- ie., the ability to modify the environment as it changes in accordance with your goals.
This rough definition is inspired by work in animals to understand the role of the neocortex, and animal learning, and the role of consciousness therein. Roughly: consciousness is "perceptual and cognitive intelligence grappling with environmental change".
I am agnostic regarding that, as I don't think there is any evidence that they do not attempt to build models that are consistent representations of reality.
I am assuming, based on my own experience, they also have this "internal lightbulb" going on when they think they have built the correct model. But whether they are actually cognizant of it (self-aware), I have no idea. (I guess what I am saying is that understanding and self-awareness are two different things.)
I'm not even talking about self-awareness. I'd be happy to raise the bar to that level when (, if) we have mice-level AI.
However the bar is way below that at the moment, and masquerading as "intelligence".
Current machine learning (ie., mere statistical) approaches to AI, that do not explicitly aim to dynamically model environments/goals/behaviour/etc., aren't even meeting an extremely minimal notion of intelligence.
We have at the moment "smart rocks". Electrical current "tumbles down" a "digital mountain" and we all it's path "smart" because it has useful outcomes. Equally, a rock rolling down a hill finds an optimal path -- it aint "smart".
We should look at what the rock does when you start adpating its environment: eg., create a little dip in the mountain side; it gets trapped. A mouse doesnt get trapped in a dip, it continues to explore -- why?
Because animal behaviour is inherently exploratory of the enviornment. A mouse doesnt "solve" a maze, it intelligently navigates it -- so that when unexpected change occurs, it isn't "broken".
At the moment, all AI systems radically break when such changes occur -- because they are statistically trained on mere data. They arent dynamically model building. They aren't in an environment. They're just rocks rolling down a hill.
There is a lot of evidence that they do though - wolves splitting the pack to ambush the tired deer at the end of the valley, chimps and corvids using tools and water displacement to achieve goals, whales bubble fishing in teams.
Very good observation, although I'd say this is still just understanding at the micro-level. A lot of what is going on in communication between people depends just as much on what hasn't been said, what would normally be said in this situation, having an idea of what the situation is in the first place, what was said recently or the last time you interacted with this person (which could potentially be a very long time ago), etc. I do believe that a lot or all of this can be posed as CSPs though.
On my reading list is "The proper treatment of events", a book which "studies the semantics of tense and aspect" within a formal framework of constraint logic programming[1]. There is other similar work in this area, like "Good-enough parsing, Whenever possible interpretation:a constraint-based model of sentence comprehension"[2].
Question: What is an example of a bird?
Answer: An egret.
Question: What is another example?
Answer: Canaries.
Seems to do fine. I don't really have a stop though, so it goes on making up new questions on it's own. Make of it what you will. Very few of the answers are correct or even coherent enough to be correct:
https://hastebin.com/agululiqif.txt
I do like this one though:
Question: Who is the inventor of the English ham?
Answer: Poor old Francis Bacon.
I think you're right. The word "to understand" has two meanings. In the narrow sense, it's the feeling that we have when we "get it", that is we think that we have built a correct model of reality and it passes the logic consistency check (which is verified by being able to give an example or counterexample). In the broad sense, it means to build and apply correct models of reality.
Above, I am talking in the narrow sense. So the fact that the model itself is wrong shouldn't be an issue. But in the broad sense, we could say that understanding is ability to convert between intensional and extensional (ostensive) representations (models) of the world. Finding an example from intensional representation is just one task that is required.
But perhaps I wasn't clear, the study doesn't say this, but it was rather my own experience with the BS sentences in that study that led me to the observation that they have an empty set of examples if we take them as a constraint satisfaction problem of sorts.
Also to be able to manipulate and use the conceptual realisation ? "I could use this twig (but not that twig) as a hook for ants if I strip the bark off and turn it when I hold it i the hole, then I can eat the ants"
On the one hand the quote by Edsger Dijkstra comes to mind. "The question of whether machines can think is about as relevant as the question of whether submarines can swim." We are hardwired to attribute great significance to what happens both in our own head and that of other people.
On the other hand, machines still perform actions that one could call 'stupid'. When alphago was losing in the fourth match against Lee Sedol it would play 'stupid' moves. These were, for instance, trivial threads that any somewhat accomplished amateur go player would recognize in an instant and answer correctly.
Humans, and also animals, have a hierarchy in their understanding of things. This maps on brain structure too. Evolution has added layers to the brain while keeping the existing structure. In this layered structure the lower parts are faster and more accurate but not as sophisticated. Stupidity arises because of a lack of layeredness so when the goal of winning the game is thwarted the top layer doesn't have any useful thing to do anymore and it falls back on a layer behind that. For alphago pretty much the only layer behind its very strong go engine is the rules of go. So, even when it is losing it will never play an illegal move but it will do otherwise trivially stupid things. For humans there is a layer between these things that prevents them from doing useless stuff. For living entities this is essential for survival. You can be forgetful of your dentist appointment but it is not possible to forget to let your heart beat. It seems that this problem could be mended by putting layers between the top level algorithm and most basic hardware level such that stupid stuff is preempted.
> When alphago was losing in the fourth match against Lee Sedol it would play 'stupid' moves. These were, for instance, trivial threads that any somewhat accomplished amateur go player would recognize in an instant and answer correctly.
I think this behavior is less 'stupid' than it appears. When human beings play Go, the points matter even to the loser, and everyone goes home when it is over. There is life outside of Go. To Alpha Go, Go is it's entire universe. Part of the way it was trained was competing against other instances of itself, a sort of Thunderdome where the loser doesn't get to continue existing, and doesn't contribute to future generations. To Alpha Go, defeat is death. The behavior we observe when losing is nigh-certain has a human equivalent, we call it desperation. Alpha Go is trying moves that can only possibly work if the opponent makes a catastrophic blunder, which is incredibly unlikely, but it's the only shot it has.
> When I ask Google “Who did IBM’s Deep Blue system defeat?” and it gives me an infobox with the answer “Kasparov” in big letters, it has correctly understood my question. Of course this understanding is limited. If I follow up my question to Google with “When?”, it gives me the dictionary definition of “when” — it doesn’t interpret my question as part of a dialogue.
Google Search doesn't, but Google Assistant does. I posed the exact queries suggested by the article and the second query of simply the word "when" did give the correct answer (May 11 1997).
I remember that when my friend got a Google Home almost 2 years ago, I was asking it some questions to explore the limitations. I asked about a certain restaurant chain, and it gave me the information, but then I asked "is there one near me". It listed all places with "one" in the name near me.
I wonder if now it would correctly take the previous context into account. Google has been working a lot on improving their search and assistants to be "conversational". [1] looks like one of the results of this endevour.
That example seems pretty unrelated to what I would think of as “understanding.” That’s more just a feature request for Siri.
It’s like saying “my calculator lets me type ’1 + 2 =’ and gives me the answer ‘3,’ so it seems to understand that question, but when I look at the calculator I see there’s no ‘sqrt’ button that would show me the square root of 3.”
The fact that my basic calculator doesn’t have a “sqrt” button is pretty irrelevant to how well it “understands” how to add two numbers together.
Your basic calculator still has a concept of context, though. If you go '1 + 2 =' then it will give you '3', and if you press '/ 2 =' then it will give you '1.5'. It 'remembers what you were talking about' within its very limited scope.
I think what they were trying to get at is that understanding is stateful.
I don't remember where I first saw it, but the best definition of "understanding" I've seen is "being able to encode and compress".
For example, imagine a system that has as input the picture of a human face in RAW format. If the system runs the picture through JPEG compression, for example, and returns something substantially smaller, it has shown some understanding of the input (color, spatial repetition, etc).
A more advanced system, with more understanding, may recognize it as a human face, and convert it to a template like the ones used for facial recognition. It doesn't care about individual pixels anymore, or the lighting, just general features of faces. It understands faces.
An even more advanced system may recognize the specific person and compress the whole thing to a few bits.
I would say that an OCR scanner understands the alphabet and how text is laid out, GPT-2 understands the relationship between words and how text is written. And a physics simulator understands basic physics because it can approximately compress a sequence of object movements into only initial conditions and small corrections.
Lossy compression makes this concept non-trivial to measure, but it's still a world's away from the normal philosophical arguments.
> Speaking as a psychologist, I’m flabbergasted by claims that the decisions of algorithms are opaque while the decisions of people are transparent. I’ve spent half my life at it and I still have limited success understanding human decisions. - Jean-François Bonnefon’s tweet (as quoted in https://p.migdal.pl/2019/07/15/human-machine-learning-motiva...)
The advantage of humans is that we have a building bullshit generator.
If someone ask why you like ice-cream, you can tell a nice story about the hot summers during your childhood, but the reality is that sugar and fat are very useful.
If a the autopilot of a Tesla hit someone, the error report is "Fatal error 0xDEADBEEF: coefficient 742 > 812".
If a person hit someone the explanation is "It was dark and near a curve. I was texting that is totally safe. I got distracted by reindeer nearby. And I snoozed and was thinking about reaching a handkerchief".
To be gadflyish do humans even truly understand or do they just claim they do because they had the observations roughly encoded from what they have been taught? Teachings which themselves often include unfounded assumptions or outright superstition.
Human understanding has been wrong often enough, missing enough crucial context to be dangerously hillariously wrong even amongst the "experts" of the day who came closest.
The isn't some epistemological nilhism but to point out that understanding is incomplete for everyone and just because a given intelligence subset doesn't match with our assumptions doesn't mean it is wrong - although it also isn't always right.
I think getting near human level for NLP understanding means be being able to visualize and combine all of the dynamic systems that language represents. I mean it's obvious that you can get pretty far just by processing a lot of text, but there is a limit. Some information about the way things work just is not encoded very well in text the way it is in video input. So you need to be able to do a sort of physics simulation for starters. Except it can't just be physics, because there are a lot of patterns that occur that you need to be able to call up and manipulate or combine that are not just plain physics. These patterns are not represented in text.
There are projects doing video and text understanding. I think the trick to efficient generalization is to have the representations properly factored out somehow. Maybe things like capsule networks will help. Although that my guess is that to get really sort of componentized efficient understanding neural networks are not going to be the most effective way.
The proposal in the article is to define "understanding" and work towards testable satisfaction of the definition.
This sounds a bit like a studying for a test taking. What if we made a definition and then worked successfully to reach the state when, according to this definition, the system "understands". Can we expect to be satisfied with the result in general, outside of the definition?
The definition of understanding could be tricky, as history suggests. Other than "to understand is to translate into a form which is suitable for some use", there could be many definitions. Article itself brings examples of chess playing or truck driving which were considered good indicators, yet failed to satisfy us in some ways.
Maybe we should just keep redefining "understanding" as good as we can today, and changing it if needed, and work trying to create a system "good", not necessarily "passing the test"?
OK, wow, the old guard sure knows how to write sensibly. This is a great
article.
But I have to disagree with this (because of course I do):
>> For example, when I tell Siri “Call Carol” and it dials the correct number,
you will have a hard time convincing me that Siri did not understand my
request.
That is a very common-sense and down-to-earth non-definition of intelligence:
how can an entity that is answering a question correctly not "understand" the
question?
I am going
to quote Richard Feynman who encountered an example of this "how":
After a lot of investigation, I finally figured out that the students had
memorized everything, but they didn’t know what anything meant. When they
heard “light that is reflected from a medium with an index,” they didn’t know
that it meant a material such as water. They didn’t know that the “direction
of the light” is the direction in which you see something when you’re looking
at it, and so on. Everything was entirely memorized, yet nothing had been
translated into meaningful words. So if I asked, “What is Brewster’s Angle?”
I’m going into the computer with the right keywords. But if I say, “Look at
the water,” nothing happens – they don’t have anything under “Look at the
water”!
In this (in?) famous passage Feynman is arguing that students of physics that
he met in Brazil didn't know physics, even though they had memorised physics
textbooks.
Feynman doesn't talk about "understanding". Rather he talks about "knowing" a
subject. But his is also a very straight-forward definition of knowing: you
can tell whether someone knows a subject if you ask them many questions from
different angles and find that they can only answer the questions asked from
one single angle.
So if I follow up "Siri, call Carol" with "Siri, what is a call" and Siri
answers by calling Carol, I know that Siri doesn't know what a call is,
probably doesn't know what a Carol is, or what a call-Carol is, and so that
Siri doesn't have any understanding from a very common-sense point of view.
Not sure if this goes beyond the Chinese room argument though. Perhaps I'm
just on a diffferent side of it than Thomas Dietterich.
I think the key ingredient is 'being in the game', that means, having a body, being in an environment with a purpose. Humans are by default playing this game called 'life', we have to understand otherwise we perish, or our genes perish.
It's not about symbolic vs connectionist, or qualia, or self consciousness. It's about being in the world, acting and observing the effects of actions, and having something to win or lose as a consequence of acting. This doesn't happen when training a neural net to recognise objects in images or doing translation. It's just a static dataset, a 'dead' world.
AI until now has had a hard time simulating agents or creating real robotic bodies - it's expensive, and the system learns slowly, and it's unstable. But progress happens. Until our AI agents get real hands and feet and a purpose they can't be in the world and develop true understanding, they are more like subsystems of the brain than the whole brain. We need to close the loop with the environment for true understanding.
It certainly doesn’t understand Go as a board game humans invented as a stimulating mental exercise that became competitive enough to see whether human programmers could come up with a program that could beat any human. And whatever cultural history went along with playing Go. Certainly chess playing has been used as an analogy in the west for many activities involving strategy. This is something no computer currently understands.
Even a simple scaling down of the board from 19x19 to 9x9 has a huge effect on strategy. A circular board would probably produce something that doesn't look like Go and would confuse trained humans as well.
I'm with John Searle on the Chinese room [1] opinion, i.e. that a machine cannot be said to "understand" language even if it is able to pass the Turing Test. That is because when we say "understand", we are referring to particular kind of human experience (qualia?) that a machine simply doesn't seem to have, but animals, for example, do.
Unfortunately you have no way of determining if a machine or animal have this particular experience qualia. You cant even determine of other people beside yourself have it, which gives rise to solipsism.
It like saying that red-headed people doesn't have a soul - there is no way to disprove that assertion.
The point is not whether we can prove that machines have such an experience. The point is, most people will agree that it “seems” like machines don’t have feelings, which is itself interesting because it suggests that our intuitive definition of “understanding” is not limited to a logical set of inputs and outputs. It is a proof that when we say “he understands”, we are referring to something more than a logical answer, and whatever that “something more” is, machines don’t have it.
In puppet theater good puppeteers make the figures look like they think and feel very convincingly, even though we know they are just made out of wood. I'm sure the same can be achieved with robots.
That's the fundamental issue underlying the discussion. Pointing this out doesn't add anything. It's like saying water is wet. Yes, but how do we quantify human understanding such that we can implement it in AI? Well, we've been puzzling over that for decades and we sill haven't found an answer.
I can say that you don't have qualia and you can't prove me wrong.
Does that seem dangerous to anyone else?
I also don't see any distinction between "qualia" and "soul" other than spelling, but perhaps it's because I don't have one.
Finally, I have this question for Searle: Say you understand English. Does any specific neuron in your brain understand English? No, the larger system of neurons+neuronal connections does, so why doesn't the system of grad student+book understand Chinese?
I'm not sure where you're going with this. Are you implying that thinking about this problem is dangerous in itself? That would be absurd. We've been philosophizing about qualia or what it means to be human or whether we can even be certain anybody else shares the same experience of reality or even if they're human at all for centuries. I mean, what else does Descartes do in the process of his discourse? I don't think that's caused undue suffering in the same way that organized religion has.
All it shows is that after hundreds of years, we still don't know how to explain or quantify human consciousness.
Dangerous? No. To me it just seems to mean that “qualia” is not a particularly useful concept, particularly when discussing the capabilities of computer software.
"Dangerous" as in "Oh, you know they don't have qualia, so we can do whatever we want to them. They don't really feel pain, they just act like they do."
The issue is that we’re using a word based on human mental activity and social agreement and then applying that to a computational process in a machine, which likely leaves out part of the human experience which makes up the word understand.
It’s more accurate to say the Chinese room computes results which humans recognize as successful translation from English to Chinese. The understanding is all on the side interpreting the output.
Then what is the point in asking if a machine “understands” English and Chinese? It sounds like the question would be either completely untestable, or the answer would just be “no” by definition because we’re defining “understanding” to be “based on human mental activity.” It just doesn’t seem like a useful question if the answer can not be determined by a test such as the Chinese room thought experiment.
Well, if we asked this about Data from Star Trek, then the answer would have to be yes, or mostly yes (Data does struggle to make sense of some human behavior on the show). So then the question is what gives Data an understanding that the Chinese Room lacks?
Data participates in human society and he has a human-like body. Data also has subjective experiences, as evidence by his dream sequences in one episode. Whereas the Chinese Room is just following a bunch of rules for translation. But Data doesn't merely translate from one set of symbols to another given a large set of rules. He learns by interacting with people and his experiences as an android. From that we could say understanding is the result of an embodied social activity that the Chinese Room completely lacks. Whatever the Chinese room is said to be doing, that's not the same as understanding language.
Another way to put it is that language isn't equivalent to symbol manipulation, even though it makes use of symbols, or a least since the written word was invented.
The point of asking such a question is to prove whether or not computational processes accurately model the human brain.
Unfortunately this is a recursive question, because the only device we have for exploring the difference between a brain and a computer is our brains. Thus, I believe the Chinese room experiment is rightly composed of as a thought experiment - what other means do we have for assessing our difference from computers other than our intuition?
Do you have color, sound, taste experiences? Do you experience pain? Do you dream or imagine? Then you have quaila. If you don’t like the philosophical connotations of that term, then call it subjective experience.
It's a reasonable inference given your similar biology and behavior. Doubting your subjectivity is on the level of doubting that anything exists outside my experience, or doubting that the universe existed more than five minutes ago. I can't prove those things are true, but I have no good reason to have such doubts.
Still, it remains a philosophical problem, even more so for animals or robots like Data. That's what Ned Block called The Harder Problem of Consciousness. But here I think we just have to accept that our knowledge of others and the world lacks certainty. We trust our senses and inferences to form a reasonable view of the world, but we can never be sure.
Dangerous? Only as much as you want to swim in nihilism. Someone who claims you have qualia, a soul and a psychic commitment to astrology or whatever is more dangerous because that person has an affirmative commitment to maintaining that belief.
Qualia is generally argued by Sam Harris to be simple or reductionist elements of our human experience we can all agree humans share. Burning your finger on a hot stove and recoiling is a conscious experience every human shares.
The soul includes way more ideas and depends on who you talk to. The word has been overloaded a bunch, but generally can be said to include a higher spiritual aspect.
See my other comment in this discussion. What you experience as "understanding" is that a particular constraint satisfaction problem (roughly, the logical fragment that is supposed to be understood) has a solution and you are able to construct it.
I don't think it's possible for machines to understand. Numbers are meaningless, our human actions give them a useful function. All of the meaning a computer appears to provide is the preassigned values of layers and layers of programming work done by humans. Even today AI has a lot of human tagging and categorization that makes it useful.
The idea that a new self- sustaining meaning generation can arise out of the interlocking mechanisms of a computer is an interesting one. As we see self driven car CEOs describe some of the most advanced systems we have, requiring to be run in controlled environments and balking at the infinite complexity of real life, are we really building computer systems that are anything more than an incredibly sophisticated loop?
Well, what does it mean for humans to "understand"? Don't humans understand things by altering the state and connections of neurons in the brain? You could make the argument that the brain is also an "incredibly sophisticated loop".
My point is that humans are also highly-sophisticated, biological machines, so if you say machines cannot "understand", you are making the same claim for humans as well.
Humans also squirt fluids around in their brains. Brains as machines is one of many ways to think about humans. Humans can conceive of and move past thoughts or concepts that would cause a machine to crash. I think more ideas describe human brains than being simply machines, though that idea is useful in places.
Making the claim about what a human is in the absolute, is more about what you fill the unknown with than the nature of a human.
Understanding is the difficult question. I would argue the understanding people want out of machines is the ability to generate, use and self-manage tools and that the machine knows the tool's place or context under a human value, story or intent and adapt to the implications of that higher order. That in the most exaggerated sense would be perceived as a machine that understands, but of course people mean different things when they say that.
I get how ML works. I don't mean loops as a for(int i) loop, but the concept of a loop itself, a circle. A self-driven car with ML decision making is still bounded by some rules we will be forced to compromise on. Some people at MIT are focusing on deaths per miles driven as a safety metric to determine whether we can replace humans with ai cars and when that might happen.
But given the constraints of ML/ai you will eventually have a bounded container where an ai car can operate and where it can't. The car will be tasked with looping through that environment from job to job then back to recharge at it's base station. For all the sophistication of getting the car on the road and working it won't really be making up it's own story through the world nor will it understand the greater context of it's actions. The pattern recognition in CV is great, but it is fed by humans, so the meaning that a tree should be avoided has been initially put in by a programmer, even if the car in the moment chooses to avoid the tree by itself. The car is crunching meaningless numbers like a pipe directs water.
So when people say a machine "understands something" it can't ever really be true because all of our machines don't know what is going on in the world, they only know what numbers they see and how to behave when those numbers change. At the very bottom it's electricity looping through logic gates and that same principle is repeated all the way up to a car that loops through it's environment and comes back.
If all the humans left the planet, the car wouldn't be described as understanding the world, it'll be seen as a generic device sitting in a garage somewhere waiting for orders from a human. If you fill the earth with aliens the CV breaks not having seen aliens before, the roads get changed over time by nature, the high detail mapping it relies on fails. The cars "understanding" only exists as an outcome of electric impulse. It doesn't understand and never could. We are building more and more sophisticated loops, and I'm glad, but to think computers can understand is a doomed project. They will never "get" the values, intents and stories we put in them. Computers will forever be a labour of love is not able to regress into understanding what we mean it to be.
> A self-driven car with ML decision making is still bounded by some rules we will be forced to compromise on.
The atom in the molecules in the neurons of your brain are bounded by the laws of Physics. They can't disobey them, they are as free as the coefficient of the ML tables.
> If all the humans left the planet, the car wouldn't be described as understanding the world, it'll be seen as a generic device sitting in a garage somewhere waiting for orders from a human.
Unless some car have setup an alarm to go to pick you from work at 5pm, you are not there but it goes anyway. After some time (1 hour?) it gives up and return home to get charged and wait for the next day. The waiting time depend on the weather (if it is cold or rainy) and the battery charge and perhaps the congestion of the roads.
Once per year they go to the robot-mechanic for the anual service. They also go when a tire or something get broken. They can call the autonomous crane in case it is needed. During the repairing time, they call a replacement and send all your info and schedule, so you would not miss your appointments (in case you were still there).
The car also negotiates automatically the insurance with the company web service, and pays the registration fees. Your autonomous house pays the electricity bills. Until your bank account is empty.
If you have some money in a good investment found this can last for a long time, until your car is too old and decides to retire and buys a replacement.
We are still very far from this scenario, but it is not so difficult to imagine that a bunch of small features compose nicely.
That said, I have a small problem with the examples presented to say that already machines understand us :)
The article says 'For example, when I tell Siri “Call Carol” and it dials the correct number, you will have a hard time convincing me that Siri did not understand my request"
Let me try to take a shot at trying to explain that Siri did not "understand" your request.
Siri was waiting for a command and executed the best command that matched. Which is, make a phone call.
It did not understand what you meant because it did not take the whole environment into consideration. What if Carol was just in the other room. A human would maybe just shout "hey Carol, Thomas is asking you to come", instead of making a phone call.
If listening to a request and executing a command is understanding, then computers have been understanding us for a long time. Even without the latest advances in AI.