Would you be willing to read a novel written by a computer, or listen to a song composed by a robot? As artificial general intelligence (AGI) continues its rapid growth into all areas of society there are strong hints, aspiration even, that it could overshadow human intelligence. Are the aspects of the human mind that make people unique under threat from a technology that we have created and programmed with the aim that it will learn and adapt itself to new levels that we cannot anticipate in advance?
Let’s first consider intellectual property. Novel writing algorithms could lead to a situation when some of what is widely considered to be the best literature of the year is written by a computer. The Hoshi Shinichi Literary Award is a Japanese short story competition, out of the 1,450 entries in 2016, eleven were said to be written by ‘non-humans’. One line from a story co-authored by AGI was profound, considering the situation: “The day a computer wrote a novel. The computer, placing priority on the pursuit of its own joy, stopped working for humans.”
While one assumes that this line was written by the human author, there is still a lingering feeling that a computer could have this awareness. This would mean literary awards judges will have a significant decision to make. The question is, does it matter who created the literature? If the story is good and the composition regarded at its notional best, surely the creativity is what matters?
Computers can now create what appear to be famous paintings with impressive accuracy, conversely algorithms have been created to identify fakes by humans. Computers can flirt and write jokes, it seems that every year that goes by the differences between humans and computers decreases. The close proximity between human intelligence and AGI has led to questions about what happens when AGI eclipses us.
One of the most interesting settings for this debate will be in arts and culture, how easily accepted will AGI authors and musicians be accepted or will they be rejected outright. The art critic, Martin Gayford, wrote, “The unresolved questions about machine art are, first, what its potential is and, second, whether—irrespective of the quality of the work produced—it can truly be described as ‘creative’ or ‘imaginative’.”
If people decide that all work by AGI is not creative because it came from a computer programmed by a human, then a world of artistic possibilities will be lost. Surely though this means that it is at its ‘heart’ still human, as we created the algorithm. Will customers care that what they are reading is created by a machine, using vastly different techniques than people use. Perhaps more interesting then, is to ask who would gain credit for the work, the computer or the human(s) who created the AGI.
Mark Zuckerberg has predicted that in a decade AGI will start outperforming humans, with Facebook trying to build AGI with better perception than people. This relates to some of the basic human senses like seeing and hearing. I venture that in many (highly specific) scenarios this is already the case. The main way AGI will redefine human intelligence is by making a lot of it irrelevant, many workers will be looking over their shoulders, not at robots, but computers.
If AGI is more perceptive, that does not mean that it is better, Zuckerberg made sure to calm the human population by confirming that computers will not be “thinking or generally better [than humans]”. For example, a computer may be programmed to possess the ability to beat a person at chess, but the computer cannot stop the game and write a letter about its style of play. The same computer cannot beat a human at backgammon either, for this to happen it would need to be reprogrammed by a human first.
What we are waiting for, and seems a long way off, is a computer programmed to learn by itself from basic principles as we do. Given, as we (humans/animals) are, some critical base pieces of hardware and algorithmic software. Then put into the wild, with teachers (humans to start with) and allowed to grow from baby steps; will the computer truly be in a position to go beyond human capability, learning and mastering chess, then moving to the backgammon table and mastering that too?
One of the most unique creations of human intelligence, written language, must also be adopted by AGI, as you’ve seen above in many respects it has already started. There have been reports of AGI programmes being shut down because the AGI bots created their own language, these stories have been proven to be wide of the mark and somewhat embellished. However, AGI creating its own language could be the key to its success as a new kind of intelligence in the future, this seemingly scary prospect could be the point where AGI overtakes human intelligence.
AGI with its own ‘self-taught’ method of processing information will not only be seen as the time it took over human intelligence, but also the time where people fear how safe it is to be continually developing AGI.
One of the most uncomfortable aspects of AGI for some humans is what appears to make us unique will no longer be the case. Humans propose that they could understand aliens from another planet creating art and music, but with it coming from technology that is controlled and programmed by humans will leave many people feeling uneasy.
Taking the positive stance, another question we could be asking, is what is it that we really fear? If AGI does redefine human intelligence, surely that will also allow humans to reach levels they otherwise would never have attained. It could be seen as a new era for humanity where we finally transcend hate and inequality through common purpose and greater understanding of ourselves.