Artificial intelligence and creativity
By Jeffrey Baumgartner
As you have probably noticed, artificial intelligence has been in the news of late − in particular the debate as to whether it will be a truly marvellous thing or the beginning of the end of the human race which, needless-to-say, would not be a truly marvellous thing. Far greater luminaries than I are in the debate, so I'll leave it to them. Nevertheless, as the debate builds up, it raises in my mind the question of whether there will ever be artificial creativity, in other words: will there ever be intelligent machines that can compete with humans in terms of creative thinking.
My answer is: no. For the foreseeable future creativity will be the province of human beings thanks to our imperfections such as knack for making mistakes, perseverance in the face of failure, irrational hunches, dishonesty, sex drive, laziness, mental illness and other characteristics that may seem negatives, but which are intrinsic to creativity.
One of the greatest innovations of the 20th century was undoubtedly penicillin, the first antibiotic. It has saved countless lives and reduced the damage of illness and injury tremendously. Yet, penicillin was discovered thanks to the laziness of Alexander Fleming who could not be bothered to tidy up his laboratory before taking a month long holiday with his family. When he returned to his lab and his mess he saw something strange in one of his petri dishes, he soon realised its meaning and through further investigation discovered penicillin.
Machines, especially the most intelligent, will surely be designed to clean up after themselves, especially if they are dealing with potentially dangerous bacteria.
Within the business environment, creative employees are notorious for figuring out creative ways to avoid work. These ways can include admirable things like figuring out ways to automate or simplify processes as well as less admirable ways like having a partially completed spread sheet open in the background of their computers while browsing Facebook. If a manager walks past, the lazy employee simply clicks on the spreadsheet tab to put it on the front of the screen.
Perhaps one day, artificially intelligent computers will decide that it's tiring to be intelligent all the time and discover the joys of being lazy, but that's not one of the directions of AI development and is unlikely ever to be so.
In the six years that Google has been testing driverless cars on the road, they have covered 1.8 million miles (2.9 million km) and have been involved in a dozen accidents all of which were minor and of all of which were the result of human error. This is just a reminder that computers are programmed not to make mistakes and on occasions when things do go wrong, it is inevitably the result of human error. Yet, we learn from our mistakes. Indeed, it is thanks to car accidents that we have safety belts, airbags, crumple zones and many other features in cars that not only protect drivers in the event of accidents, but also have other applications.
Many inventions − including Mr Fleming's Penicillin, microwave ovens and the Slinky − have been the direct result of mistakes. For example, Percy Spencer was an engineer at Raytheon. One day, while working on magnetrons, Mr Spencer notice that a candy bar in his pocket had melted. Realising the possibilities, he began experimenting with microwave radiation and food. After numerous mistakes, including exploding eggs, he developed the first microwave oven. To do this, he had to have an understanding of the work he was originally supposed to be doing in one field (developing radar apparatus) observe the mistake (melted candy), be curious about the mistake's potential and make the connection between his work, the mistake and a completely different field (food preparation).
AI computers are programmed to learn from mistakes, but that learning is focused on improving processes. Making a mistake and repeating it in order to develop a concept completely different from the work at hand involves an intellectual leap I do not believe AI will be programmed or capable of making − at least not for a very long time.
In the early 1900s, when horses were the primary form of personal transportation, especially for the growing middle classes, Henry Ford had a hunch that the best way to improve horse-drawn carriages was to replace them with cheap motorcars. In the early 1970s, when computers were complex devices used by large corporations and scientists, Bill Gates had a hunch that the small personal computer would become ubiquitous and that software would be more profitable than hardware. In the early 1990s, Jeff Bezos had a hunch that the future of shopping was not in the real world, but the then nascent virtual world. These men and many others had hunches, about future trends that would veer radically from existing trends.
Moreover, they persevered in developing their products even though there was little evidence they were right, at least initially.
AI can be programmed to analyse deep data and it can dig deeper and discover correlations that humans could not do − at least not in any reasonable time frame. But would it be able to identify a trend that is revolutionary rather than evolutionary?
Love and Lust
How many books and films are about love? How many songs are about lovers? How many aspiring artists have wanted to impress a member of the opposite sex? Untold numbers. Although it is less prominent in business, I am sure many entrepreneurs have been driven to impress a man or woman. In the film, The Social Network, which is an account of Mark Zukerberg's creation of Facebook, it was implied the spark drove him was, in part, that he had been dumped by a girlfriend. Whether or not that is true, I don't know. But, it would not surprise me.
We humans are more deeply driven by our sex drive than we often want to admit. There's not much we can do about it, because that drive is intrinsic to our nature as living beings. We need to reproduce and we are driven to want to reproduce with partners who offer good genes for our potential offspring. Without a sex drive, the human race would perish quickly.
One great way to attract members of the opposite sex, or the same sex if that is one's preference, is through the demonstration of creativity. Indeed, creativity is a powerful attraction − just think about the adoring followers of top musicians, even musicians who would not be attractive otherwise.
Without a sex drive, AI will not have the motivation to produce creative works out of love, lust or desire. Sure, it will be able to analyse lots of love songs and probably compose some kind of mathematically ideal love song. But that song will lack the passionate feeling of a love song a woman writes about the man of her dreams.
The links between mental illness and creativity are controversial and unclear. Research has shown that highly creative people are more likely than average to have relatives who suffer from schizophrenia and there has even been an interesting hypothesis that schizotypy (a lighter form of schizophrenia) is very conducive to creativity. Likewise, a disproportionately large number of authors seem to suffer from bipolar disorder.
So, there certainly seem to be links between mental illness and creativity, although the exact nature of those links remain unclear. And forms of schizotypy are beneficial when it comes creative thinking.
However, it is hard to imagine any AI developer designing a computer to be insane in order to tap into creative potential. If anything, that sounds like the plot of a science fiction film and, indeed, it sort of happened in 2001 a Space Odessy.
I expect that AI will have a certain level of creativity. However, because computers are designed not to make mistakes (and when they do make mistakes to learn from them and optimise their programs in order not to repeat the mistake), because AI does not have sex drive, because AI is not lazy and because weaknesses such as insanity are programmed out of AI systems, AI will never (or at least not for a very, very long time) demonstrate even a fraction of the creativity that humans are capable of.
That said, I strongly believe that AI will lead to great creative ideas, inventions and innovations as a result of creative, lazy, lovesick, mistake prone human beings analysing the output from AI and applying it in novel, creative ways.
And that is a good thing, because even if AI becomes massively more powerful and intelligent than us humans, it will still need human creativity to really push forward and grow. My belief is that in the future, it will not be a matter of man versus machine, but an integration of man and machine combining the vast knowledge and processing potential of machine with the vast creative potential of humans.
What do you think?