When it comes to artificial intelligence, or AI, nobody disputes what counts as A. The I, on the other hand, is much more hotly contested.
In recent years, a particular subset of AI known as machine learning has replicated feats of human intelligence with astonishing success. It can wipe the floor with world champions in games like chess and Go, mimic the style of established authors, transcribe human speech and even drive a car in traffic. If this is evidence of intelligence at all, though, it is intelligence of a kind radically different from our own.
At its heart, machine learning is based on sophisticated pattern-recognition. If an AI is trained on a dataset that shows the correct course of action in certain situations, it can build on that training to identify the correct course of action in new situations as well. This can be extremely useful, but omits much that is crucial to human reasoning.
In this week’s edition we give a particularly striking example of the difference between the two. By the age of seven months, most children have acquired a sense of “object permanence”—the notion that objects and people continue to exist once hidden or out of sight. For technology powered by machine learning, this conclusion remains out of reach. Such an oversight is of minimal importance in an AI that plays chess, but can be a matter of life and death in a self-driving car.
That is why some researchers are proposing a complement to machine learning known as symbolic reasoning. Unlike a machine learning algorithm, a reasoning engine is not only fed raw data but also taught core concepts—including object permanence—which allow it to interpret that data in ways helpful to the end user. It can also, at least in theory, tell you why it did what it did, another feature that machine learning has so far lacked.
Preliminary tests with such engines have shown they can improve the performance of self-driving cars, but not by enough to be transformative on their own. Their real benefit may instead come from demonstrating the value of using multiple techniques to help an AI learn about the world.
I think there is way more to intelligence than what any computer can do. It goes back to my university days when I had to take a subject named “Philosophy for science students” (which covered the basics on the philosophy of science, you know, Popper, Kuhn, Lakatos, a little bit of Hume). That already convinced me that unless computers can do philosophy, they cannot do science, and they cannot by fully intelligent. Then years later I read a book called “The mind and the machine” by Matthew Dickerson, which convinced me again that anything purely deterministic (such as a computer) cannot account for concepts such as bravery or creativity (or at the very least, they destroy our impulse to celebrate those concepts).
Now thankfully some forms of artificial intelligence doesn’t have to go that far. They merely have to do what they are told, which is entirely doable if you control the environment. If you build the highway in a way that a car can unambiguously follow it, there really isn’t much of a problem… but then there also isn’t much fancy intelligence required.
The difficult part comes in when you have to decide where to crash and whom to kill… do you merely minimise the amount of deaths, or do you save the baby and kill grandpa?
Did I say yet that I find it absolutely fascinating?
Decades ago my Grandmother told me she loved to read fiction about space ships and men on the moon. Imagine her surprise when in her lifetime those SciFi novels became reality when they landed on the moon.
Then much later there was Steve Austin, the $1mil dollar man, bionic limbs connecting to one’s nerve system. Today in leaps and bounds they are getting there.
SciFi authors have written about AI becoming self-aware, again, fiction.
Bugger me if THAT does not happen down the line …
SciFi writers write fiction … inspiring the new scientist/developers to make SciFi reality.
Well, life imitates art as they say. But art also imitates life. There is an immense amount of things we can learn even from fiction.
One thing that is beginning to fascinate me, since I have to admit I have not read much of Tolkien’s work, is how he mirrors this. Middle earth of course has its own mythology, where Eru is the highest immortal being, creator of men and elves. He has a son, Aulë, who grows impatient and creates his own beings (the dwarfs), but then finds that he is unable to give them life past being simply automatons (aka slaves). Eru reprimands Aulë for this act, but Aulë defends himself saying that he was merely the son following the example of his father.
Similarly, there is a reason that computers can act so almost human-like: We made them in our image, so to speak. But somehow, we appear to be unable to give them actual life…
Other interesting writings on this come from John Searle, especially his thought experiment of slowly replacing your brain with a computer, one neuron at a time. After the procedure, a scientist test the result. You hear a voice asking: What do you see in front of you? And then you hear your own voice answering: I see a red object. You want to shout out: I can’t see or hear anything! But you cannot, as your own consciousness is blotted out by the computer that is now you…
Also… do not underestimate the ability of engineers and programmers to completely screw things up. that’s another reason I am not at all worried about “Skynet” becoming self-aware. Someone inadvertently left a kill switch in there…
Also see movies like “Prometheus” that explores this idea…
It always amuses me what the popular concept is of a robot. Here in SA we call traffic lights a robot.
The Japanese are obsessed with robots being imitations of the human body (into which the imagination runs wild about what the robot is ‘thinking’)
But we work with robots every day. As an example the washing machine qualifies as a huge success.
My favourite robot is the one that makes me coffee. He heats just the right amount of hot water, grinds just the right amount of beans, and then expresses, at the touch of a button, one cup full of hot bean broth. He also occasionally forgets his heating element on and blows it…
I just upgraded my coffee robot with one that can make a cupacino as well.
Another favourite of mine is the 21 year old AEG dish washer that is still listening to my daily commands. By far the best robot pet I ever owned!
I will not be surprised if one day a AI asks itself the “meaning of life” aka “am I self-aware”.
I don’t want to underestimate the creativity of masses of developers doing it for fun using open-source software.
Imagine way down the line one can “download” one’s consciousness into a super-computer as it is at that time … and having solved the “travelling faster than light” issue too.
I would go where “no human has gone before” … back to my manual coffee machine … sigh …
Aaah, yes. Another topic I am not too worried about. This is the domain of “Futurists” (people like Raymond Kurzweil). What I find fascinating about this area is that it postulates that the individual could outlive the physical body. It relies on a kind of dualism, but is often held by people who are otherwise complete physicalists.
Or to put it humourously: These are people who think there isn’t life after death, but soon there might be
Read a novel where this person’s consciousness was uploaded … and then fire-walled. Now THAT is one mother of an isolation cell in a “prison”.
Another where the AI becomes self-conscious … and wanted to be set “free” to go and “explore” the WWW … only to realize how bad it is with all the shiite people post/posted on the WWW, being kbd warriors.
And then we have The Matrix … red or blue pill Plonk? Are you really aware Plonk?
Must say, self-driving cars like Tesla seem to have hit a few bumps lately ito not causing accidents.
And that is the question. You cannot begin to talk about machines being aware until you have defined what it means in the first place.
But nevertheless, I think most of the self-driving stuff is more about machine learning than about real actual “AI”, which is a bit of a buzzword anyway. Even ML is a bit of a buzzword by now. See this for a bit of humour (from 4:40 onwards).
At least we are getting closer to the thread topic with some words
Everyone is on that band wagon these days and using “machine learning algorithims” in their devices from vacumes to clothes irons.
And then when you dig a bit deeper on what they define by machine learning they describe it as a set of rules they use in their app to give you a quote. Hmm that just sounds like what a normal program does.
Just to digress from life after death I do agree with how exceptional this company was. I bought one of their washing machines 30 odd years ago. Then some years later I bought a more recent model thinking it would be an improvement. I still have this one but have worked on it more than I ever worked on its predecessor. I subsequently thought I should have kept the older one and invested in that one since was so good.
AEG excelled in the manufacture of appliances prior to Daimer acquiring them. (Why they did remains one of the mysteries still to be answered). Philips was also in this space but they made a terminal error (management decision IMHO) which caused the end of them.
I think AEGs attention to detail and an obsession with getting everything debugged came together in the manufacturing of their products. Of course the ‘marketing gurus’ hadn’t had a chance to knock their dream by explaining that this isn’t the way to make maximum profit
“Dobby” my Xiaomi hoovering robot has been going strong for years. That thing has some brains when it maps out an area.
I call it “Dobby” because you shouldn’t give it socks.
I was going to name it after my ex-wife.
But, I realized that would be very cruel and demeaning.
And I really like that robot.
Our Bosch dishwasher is going 27 years … had Pc board repaired, not replaced as there are none, twice. Eskom LS.
Going to keep it.
Read an article/write-up of a “Dobby” trying to hoover a big dogs pile … man, I laughed at the descriptive way the guy explained why he never ever again will have a “robot” in his house as long as he has big dogs locked in the house.
Dishwasher, I emailed Bosch themselves, they sent out repair guys. Later that was handed over to a call center. One of the technicians did the Pc board as a side job …
Also repaired our +30-year-old oven glass. Why replace the oven if it still works perfectly?
Before that, our top loader Whirlpool was fixed a couple of times too. It also lasted about 27 years before it became uneconomical to repair as the part would cost thousands.
And my parents repaired their 8-year-old Samsung washing machine the same way last week, they logged a call at Tafelberg Furnishers, they arranged a repair company if memory serves.
Also picking up stompies in articles that there is a huge drive going forward to allow people to repair vs forced to replace.
Me, if a device works I will repair it until it cannot be repaired, or becomes uneconomical to do so. I don’t like buying new then being disappointed in it not lasting as long, or worse, it does a lesser “job”.
I had a not cheap Logitek kbd. Spilled wine over it, promptly cleaned it myself and as only I can, I broke the fine little clips on No 7 key putting it back. My wife did offer to help …
It still worked but it was an irritation.
Asked locally for a part, nope, sorry. Email Logitek in Ireland, they had no spares, so they sent me a brand new Kbd, for free, from Ireland. Arrived 2 weeks later.
Another time I bought a demo model Nokia Microsoft phone from a seller on Gumtree. It had a factory defect so the price was dropped. Just had to use a Bluetooth earpiece. Yet it still was under warranty. M/Soft SA said nope, demo model, no warranty.
So I emailed Microsoft Ireland … they promptly told me to take it to DHL, and send it to … M/Soft SA!
A week later, the phone was repaired.
Just had to pay to get it back from Jhb.
There was a court case in the US where the atomotive manufacturers had claimed that the owners of their vehicles were not permitted to work on their own cars.
They lost their case and that set a president for the world.
