The future of new technologies16 November 2016
We invite you to read an interview that we conducted during this year’s programming hackathon on 8 October at JCommerce. Marcin Szeliga is a data philosopher and independent consultant, who for 20 years now has worked with SQL Server on a professional level. During the break between lectures and the competition for database developers, at which he served as a mentor, he agreed to answer our questions about the future of information technology, careers of the future and artificial intelligence.
During the inauguration lecture of our hackathon, you said that over the next 20 years the world will change more than it did over the previous 200 years. If so, then for sure a lot of professions that we know today will disappear. So which do you think will carry on or will develop anew? Which profession is the best choice, for ourselves or our children?
First, you need to have an open mind, there is no point deluding oneself that common patterns will work in the future. I think everything will change ever faster. Perhaps those young people who have come here today to take part in the workshops and competition represent the last generation for whom driving will be a common skill. For the next generation it might just be a hobby like horseback riding for us, and most people will have cars which will drive themselves to the destination point. And that means you have to be very open to change. The most valuable professions, the most prospective, will be all the jobs that are associated with data – that’s for sure. So the job of a statistician, an analyst, but also an engineer. There will also be some professions which are strictly related to the humanities, such as a psychologist, who will lend us an ear, or the arts, so the professions which either cannot be or which we will not want to be computerized.
So then which jobs will become obsolete?
All those in which humans had to learn some rules and now apply them, such as, for example, primary care physician, lawyer and translator. These are professions in which computers have now taken over from humans. How many of us are treated by Dr. Google? In the future, instead of blindly pestering Google, we will have a computer equipped with artificial intelligence, which actually studied medicine which means that it learned on real cases – what the symptoms are, what the causes are, what happens to a person during illness, and how to treat it. And such a computer, just as it plays chess today, will someday be able to diagnose patients – perhaps even more effectively than a human doctor.
So apart from programmers, not many of us can sleep soundly?
Not necessarily, because on the other hand, there are also jobs that can be computerized, but have not been. Some time ago I worked on a project for a garbage truck that gets around without a garbage man or a driver. And indeed several such garbage trucks were produced. It was a pretty good prototype, actually went round in the morning without human involvement, and the computer and sensors controlled the mechanical arm which gathered up the trash, and the garbage truck continued on. But in the end the project was abandoned. The arm didn’t always manage to pick up the trash, or the garbage bins were not always in the right place, or someone had forgotten to put them out, or they were hidden somewhere. Probably it could have been figured out, but in the end it was definitely cheaper to just hire people to do the job. And it is quite distinctive. Once upon a time in fantasy books or science-fiction movies, it was machines or robots that twisted the screws and did the worst jobs, and people functioned as managers. But now it turns out that it is often quite the opposite. In shopping centers, logistics centers and loading bays, the computer tells the worker where to go, and a man with a speaker in his ear hears: six steps to the left, two steps to the right, the third shelf, raise your hands … These roles can be completely reversed.
It’s a bit like in the Matrix…
It could turn out in one of many ways, as computers are very good at making decisions and, increasingly often, we let them. For example, in the recent high-profile case of driverless cars, which from time to time will have to decide, for example, whether to kill a passenger or a pedestrian who forced his way onto the road.
So the most important roles will be played by the programmers, right?
Well, not quite. Because artificial intelligence is not programmed. At least not in the sense that there are sets of rules that we have input, and now the machine must abide by them. No. It works like this: at the beginning we put some data into the machine. Information, figures, some content. For example, a model of artificial intelligence, which was asked to write an essay, had previously learned the content from Wikipedia. As it read, it received the command: contribute your opinion. There were no rules which would have regulated this process in advance.
Well, but if we do not implement rules at the start, we may completely lose control of the process. As with the example of the bot which was supposed to learn how to interact with Twitter users. The result was that it turned into a racist Hitler-lover, a ‘hater’ to all around, and had to be switched off … And this driverless car which we talked about, based purely on data, would probably sacrifice the person who was older, in worse physical condition, or of less importance to society. Because such is the logic of data. But it is socially unacceptable.
I think that it will. Because data changes the way of thinking, the paradigm is changing. Today we have your beliefs, your views, so in science we have hypotheses. After they have been constructed we set out to verify them, we draw some conclusions and get to work. In this model which I am talking about, there is no initial hypothesis. Only data – text, a number, a picture. Then there’s the model, meaning that we have learned something from the data, first came the abstraction, then the generalization, we have some rules, but the rules are derived from the data. And only with these rules are we able to draw any conclusions. And because there is so much data, everything is somehow connected to each other, this model turns out to be not so terrible. We can tell it: learn from this data, then verify it using different data. And then something arises, which was missing in the approach based on one’s belief system: it conflicts with empirical evidence. This way I can very easily judge whether I was right or not.
I don’t know if you know, but it has been estimated that as many as 70-90% of scientific papers, especially in the field of medicine, are falsified. The conclusions drawn are simply untrue. Why? Because someone had their hypothesis, perhaps even somehow reached it objectively, for example, he had a group of patients, he found something noteworthy. On this basis, he developed a hypothesis, then generalized it to all of us. But without the support of the data. Because really the result came first, and the data was adjusted to fit later. Because the data can easily be juggled, if we already have our conclusions before we even start.
Okay. But people have a tendency to impose rules. Just as medicine has bioethics, which prohibits certain tests and treatments, purely because of convictions, there may be a need for such regulation in computer science, the creation of infoethics …
I don’t know what it would look like, but it would be very interesting. Perhaps it will turn out this way, but at this stage, however, it is still science fiction.
Well, yes, because we are talking about artificial intelligence, which is also still just science fiction. But now let’s focus on something that is more real. The analysis of Big Data. Is it ethical to analyze all the data on network activity, payments, GPS positions? Theoretically we are able to connect all this data to a particular person, say, a Social Security number and know almost everything about him, even the most personal, intimate details.
That’s true. And it actually happened some time ago. There’s a book, Dataclysm, or data cataclysm, about how much computers know about us. The book was written by an American who ran an online dating site. It’s a specific kind of site where you can lie about certain issues, but you can’t, for example, lie about your preferences, because you want to meet a person who you find interesting, not the opposite. So there are aspects of your privacy which you have to tell the truth about, and yet these are things that you wouldn’t want to read about yourself in the paper. It’s amazing how much people are willing to tell you about themselves after all. The analysis of such data allows us to build a complete profile. As being off-line is slowly becoming a kind of luxury, so privacy is already a luxury nowadays.
That’s why we’re starting to see regulations come into effect. The European Union ratified the ‘right to be forgotten’ last year – so these are the first steps to giving us back the rights to our data.
That’s true. But on the other hand, we know that the Americans, the Russians, the Chinese and others are also listening to and recording all the telephone calls in the world, all emails, anything you have ever said or written. There are people who analyze and archive it for some reason. The technology already allows it – the storage and processing of data is now so cheap that governments are able to do it. Being anonymous is really a luxury, but it seems to me that people don’t really want it. They are able to put a lot about themselves out there online. It’s true that if we want to enjoy the benefits of the Internet, we need to share data. The system needs to know about us. And that’s okay. However, in using our data, we must take into account the benefits and drawbacks. We sell our data for tangible benefits. And the problem lies in the fact that in reality we sell data for below its true value. Or even for a song.
Okay. We’ve gone off topic a bit. Since there are so many unknowns and so many threats, what decisions about the future can we make to minimize this risk, even a little? Even if only in a professional sense, where to start?
The key word is data. There’s more and more and there’s going to be more still. We generate it, devices generate it, and soon we will process it on an even greater, unprecedented scale. So, people who deal with this data, plus those who have an open mind, will be increasingly indispensable. Anyway, for now there is a lack of them, good professionals are always lacking in these developing fields. In a year’s time they’ll be lacking even more, and in two years more still.
But won’t it be another ‘golden direction’? Twenty years ago, parents dreamed that their children would become doctors or lawyers. Nowadays management is most fashionable. And everybody has a tertiary degree. And now they have nowhere to work. Isn’t it going to be the same all over again? What if a machine ends up taking the place of this analyst?
That’s true. After twenty years it might actually turn out that these professions related to information technology which we know today will no longer exist. Today, however, universities are not able to turn out as many graduates as the labor market needs. Information technology today is changing the world and drives development in its entirety. It changes the scientific approach; it affects all areas of the economy and human activity. Twenty years is not all that short-term a perspective. It seems to me that during these next twenty years, the outlook in terms of IT specialists won’t worsen. And what comes next? After that, we simply don’t know, nobody can predict what will happen.