Contact me for feedback or questions! I reply to everyone.
I have written before about robots taking our jobs, and what can be done to minimize the resulting suffering. Okay, let's assume we've buried capitalism and no one has to worry about material conditions anymore. Does that mean we can live happily alongside artificial intelligence? Not even close. In fact, lost jobs are the least of our worries with AI.
In 1997, the chess computer Deep Blue beat Garry Kasparov (in a best of six), the strongest chess player at that time. This was the first time a computer ever beat the world champion; and 20 years later, this match is still talked about. Today, you can download a chess engine for your phone that will easily defeat any human player on Earth.
But how do the chess programs work? Mostly by analyzing millions of chess piece positions per second, which no human could ever do. This, of course, cannot be truly called "intelligence", but it shows that brute force can sometimes generate results better than actual intelligence. So is that all? Is "artificial intelligence" simply limited to being fast?
In December 6, 2017, Google's chess program AlphaZero played 100 games against Stockfish (the strongest chess program at the time), won 28 of them and didn't lose a single one. https://www.chess.com/news/view/google-s-alphazero-destroys-stockfish-in-100-game-match (archive). However, there was a big difference between Google's program and the others. Usually, chess engines use algorithms and move databases created by chess experts over many years (or decades). AlphaZero, however, learned to play chess knowing only the rules. It was analyzing only 80000 positions per second compared to the 70 million of Stockfish - therefore, it was choosing its moves more like a human, than a program, would. Of course, the report was exaggerated by marketing - Stockfish couldn't use its opening book, which are essential for these programs. But still, the results (with little human input) are very impressive.
But chess is not the only game where computers outperform the humans. In 2016, Google's program AlphaGo played the 18-time Go world champion Lee Sedol in a best of five. Lee thought he would
destroy AlphaGo 5-0, but in the end won only one game out of five. One commentator said, AlphaGo won so convincingly as to remove all doubt about its strength from the
minds of experienced players. In fact, it played so well that it was almost scary
. Some time later, AlphaGo also beat another world champion, Ke Jie, 3-0. Go has usually
been considered unsuited for computer play, requiring more human intuition and creativity than chess. But as we can see, machine learning (similar to what AlphaZero did in chess) worked here as
well.
Murray Campbell, one of Deep Blue's developers, called AlphaGo's victory the end of an era... board games are more or less done and it's time to move on.
What about other games? For
example, first person shooters? There has been success in making AI play those as well, using only what can be seen on the screen. https://arxiv.org/abs/1609.05521 (archive) We show that the proposed architecture
substantially outperforms built-in AI agents of the game as well as humans in deathmatch scenarios.
Okay, but there are just games - what about real life? Not so fast. For people such as Garry Kasparov, Lee Sedol and Ke Jie, those games are in fact their "real life". And a computer - just
like that - beat them in their "real life". How do you think they felt? Kasparov, after losing to Deep Blue, accused IBM of cheating (as in, that a human was helping the computer). IBM refused
to give him the logs that he asked for, so he considered that as a confirmation that they were in fact cheating. How about the Go games? China didn't allow Ke Jie's defeat to be seen on
television, because that would hurt the national pride of a state which holds Go close to its heart.
https://www.theguardian.com/technology/2017/may/24/china-censored-googles-alphago-match-against-worlds-best-go-player (archive). So, as
you can see, AI has huge effects on people's feelings, their pride, the relevance of their life's work. But really, what about something other than games?
tests by asking professionals to listen to Aiva’s pieces – and so far none of them were able to tell that they were composed by an AI.
AARON paints whatever it wants within the confines of its knowledge, driven only by its limited version of an imagination and not at any point instructed explicitly by Cohen or anybody else.
An empty heart completely suffocating,
A body full of flesh and blood devoured,
Forever burning like a candle blazing,
Surrounded by the flames of being showered.
You can test it AI here: http://52.24.230.241/poem/advance/ (archive)
For example football, where the InStat service provides detailed data for 800 clubs all over the world. That data is then, of course, used to decide transfers, during matches and in training young players. So, we can say that sports are not fully human anymore, since people don't play according to their talents, but data from AI.
I could go on, but as you can see, the AI is intruding into what has usually been human territory - until now. What's the future of humanity? Well, there are only two ways this can end. Either we destroy the AI and the whole technological system, or we will have to merge with it - since it will do everything better than humans. I mean, what would be the point of having a human brain, if a computer can outperform it? You want to write a poem - well, AI is better at that, so why not put the poem-writing AI into your brain? Or, you need to translate something? The Live Translation Chip in your brain will do that instantly. Driving somewhere? Your self-driving car will find the shortest path! Going shopping? Your intelligent fridge will know what you need, automatically buy it and the Amazon drone will bring it to your house. Talking to people? Language is so primitive - just send the information directly to someone's brain (or a robot!), over the internet. Learning? Just hook a person up to Wikipedia! Sports? Forget about training and individuality - AI chip in the brain will tell you the statistically best move. Eating? AI will detect the nutrients we need, and inject them into our blood (why have a mouth then?). What about birth? AI will detect the child's "bad traits" and remove them (variety be damned). There are many other effects that AI will have, most of them cannot be predicted now; but one thing is for sure - in the next decades, the human will become increasingly irrelevant. That is, unless we choose to destroy the technological system, and preserve humanity.
What will surely NOT happen, though, is the continuation of the current system with humans living like nothing has happened - because it will be impossible to deny the effects of AI. It will be everywhere; and will intrude into your basic everyday tasks. And then, the thought that humans are becoming replaced will start appearing into people's minds. They will ask themselves, "what's the point of us, if the AI is doing everything? Are we being reduced to laboratory rats, being fed and entertained, but having no decision making power of our own?". Eventually, the AI will also do the tasks that governments are doing these days, further disempowering us. People have many psychological needs, such as communicating with others, or the feeling of pride I've talked about earlier. One of the most important of those is the feeling of control, and AI will mostly take that away. In the end, we will either have to modify our needs (probably the AI would do that too...) - ceasing to be human - or destroy the technological system - and the decision has to be made soon.