As we are still in the middle of the arctic blast hitting the mid-west, south and north-eastern parts of the US, I noticed that my biggest fear is not hacking through the ice and snow to just make the sidewalk somewhat visible again, but if – or when – the power will fail. I myself, as many millions of fellow citizens over here, have been breaking our backs almost every day for the last 4 days to get the snow out of the way. Forget about the salt… it is not available anymore. We are in luck that we have a back alley and the trucks usually start their salt dispensing behind our house, leaving heaps of extra salt, which allows us gladly to make the roads and sidewalk slightly less slippery.
Then, just a couple of days ago, I settled in at night in the comfort of our home, and decided to run Terminator 2 one more time. Ah, the paranoia of computers being able to go online themselves and destroy the human population. Picture this, that would have happened in 1997. The movie is from 1991. Just think about it, if right now those computers would have been running on tech from 1997. Just to know… running Windows98? Really? They would come crashing down out of the sky without interfering. I am still amazed that non of the story-lines were able to just have some professional hacking team get in there. Somehow we are still not able to keep our computers secure.
Anyway, what I am getting to is that the whole thing, the fright of computers becoming so smart – especially at the end of the last millennium – that it is still not anywhere near that kind of intelligence. Oh, there is definitely some incredible process being made with working on artificial intelligence, but the standard rule that a computer can only be so smart as how it is programmed, still stands. Let’s be honest though, that still means that a computer can be smarter than a huge amount of the population 😉
I hear a lot of talking about ‘smart systems’ in my line of work. Every client wants a ‘smart system’ and to be honest, it never exceeds just having a project that can make a simple A or B decision, based on behavior. That is not smart. My ‘Smartphone’ is not smart. I like my phone, I am an extremely happy WindowsPhone user, and love it. But it is not smart. It doesn’t do work for me, it simply reminds me what to do.
A colleague told me last week that the next storm to hit on Thursday would not be so big, only about 1-2 inches of snow. That was correct, if you looked into the daily forecast. I told him to look into the weather-warnings. There it was mentioned that those 1-2 inches of snow were actually 14-18 inches. The standard weather system is not able to handle extremes all by itself.
One of the things that makes it so very hard for real smart systems to be developed, especially in AI, is that development is these days a very clustered and individual process. There is not real teamwork anymore in technical development. Innovation is very hard to find. But in this generation every creative developer wants to have their own ideas shine, which usually is only a very small element. Look at the invention of apps. They are a blessing and a curse on their own. Look at your phone or tablet, and all the fragmented tools you have on there right now. There are no ‘suites’ on there – a set of tools working extremely well together – because everyone is just building one little thing. It makes your phone or tablet a toolbox, but not the one amazing thing to have.
Certain ‘suites’ are showing already how it can be done well, but then, in their size of a digital behemoth, they are difficult to innovate on their own. Adobe’s Creative Suites and the Creative Cloud for one, and Microsoft Office as another one. And, a lot of the security suites as well. Especially the latter has to become intelligent, because a security suite needs to be aware to recognize certain not-user initiated behavior, and the user should be able to completely rely on it. But now, in the time of the big exposures of the NSA having their hands in so many elements, can we even be sure those security suites do not have all these back-doors available as well? Or in other words, do we trust the systems to do what we want to rely on?
The Adobe or Office tools work well together, but don’t do things really on their own. They are tools, but not really smart. A system like Pandora, or Spotify or XBox Music are not smart in providing me music based on what other people liked that also liked these songs; it is a very simple setting that people put into it. Steam just introduced their Game Tag system, which allows you as a player to classify your games, and it will be easier for steam to find similar games that you might like.
But it is still not smart. When I talk about ‘smart’ computing I want J.A.R.V.I.S. from Iron Man, I want the system from Minority Report, HAL 9000 for crying out loud. The question is that calling a smart phone smart does that shows the dumbness of our minds? I mean, if the weather systems are not able to hook up their extreme weather forecast right into their regular weather forecast, how far along are we then?
A computer is still not a ‘help’. It is still the old fashioned calculator. It only provides you with feedback based on the things you put into it. It is getting better though, but the whole concept of a computer actually learning from your behavior is still not really utilized. A computer is still not able to analyze you without actively asking you direct questions, and with that, a computer is not able to analyze you at all yet. If I have to ask you the question if you like tomato ketchup, it is not analyzing your taste of tomato ketchup, it is plainly asking you the question. You could lie to me and mention that you don’t like tomato ketchup, while you actually do. I could also, to get my answer, observe you, check your behavior. If you poor tomato ketchup on your plate and use it, I might very well make the conclusion that you like tomato ketchup. And if you eat burgers everytime you go out to diner, I might interpolate that information that you might like tomato ketchup on your burgers. I didn’t have to ask you any question at all. If I hum along with a song in my car, or hit repeat a couple of times, I don’t have to say that I like a song or not. If I skip a song, it is clear I am – at that time – not in the mood for it. No questions asked. You as a user still give feedback, but it is a different way of handling it. No ratings, just behavior.
But then, isn’t that looking into behavior that we are all so very afraid of these days? That we want our privacy, don’t want the idea that a company or system is looking to much into our behavior? What if your system knew so very well what kind of pornography you like and it could be hacked into some way – because, let’s be honest, none of us ever looks at pornography – right? We simply do not want that to come out.
And that is where the fate of our smart system is right now… we will not allow it to get smart. It means giving up control. Even if developers would stick their minds together, come up with amazing concepts and technologies, we might not want it at all. We want a system that helps us, but only how we want it. Maybe we are not ready for smart computers yet. And that is where CyberDyne systems never stood a chance in real life, especially not in 1997.
By the way, even if it did… bringing the whole story back to our arctic weather in this part of the world right now… with such a massive loss of power, they would not have stand a chance in any way. My computer still goes blank once the power goes out. And somehow I don’t see a T-800 climb up a ladder and fix a power line in a blistering storm.
Would be a sight to see though.