The leap from artificial intelligence to artificial curiosity is a leap across the Rubicon into a new world of synthetic consciousness.
Artificial Intelligence is an old idea, visited often by philosophers over the centuries. To researchers, programmers, and self professed nerds, the ultimate achievement in the field of AI has always been to create a machine interface, intuitive and adaptable enough to provide human users with a natural experience. Technologically, we are witnessing breakthroughs at an unprecedented pace, but philosophically we are still grappling with the meaning of intelligence, artificial or otherwise.
Blame the science fiction genre if you think an artificial intelligence interface should mimic a human personality. Most of us have been primed on the finer points of AI by Hollywood; the HAL9000 computer in “2001: A Space Odyssey“, C3P0 and various droids in the “Star Wars” franchise, or the synthetic human replicants in “Blade Runner“.
Real artificial intelligence is considerably less human. IBM’s Watson became a household name by dominating its human opponents on the TV game show Jeopardy in 2011. Watson’s ability to quickly retrieve relevant information is an invaluable skill, but whether or not it qualifies as intelligent is up for debate. Intelligence, after all, is more than just matching well formed questions to existing factual answers.
In 2016, the AI team at Google’s DeepMind created an AI named AlphaGo to play the Chinese (and later, Japanese) board game, Go. Go is an abstract strategy game, unlike Jeopardy which is a question-and-answer game. Go requires players to invent their next move – while adhering to the rules of play. While Watson was programmed to quickly sift through data, AlphaGo was programmed to learn by doing. AlphaGo didn’t win its first, second, or millionth game – but it was able to remember every move it ever made in every game it ever played, building up a database of game-play scenarios that it continually refers back to; as they say, hindsight is 20/20. If you had instantaneous and flawless recall of everything you ever did, you’d be at the top of your game too.
AlphaGo has become virtually unbeatable at Go, but don’t ask it for directions to the nearest Starbucks. For that you need Siri. Apple’s voice activated virtual assistant adds a humanesque layer of functionality to Apple products with its voice recognition and verbose feedback. It can quickly retrieve information when asked in the form of question; “Hey Siri, where is the nearest Starbucks?”, or “Hey Siri, what is the largest prime number less than one million?”. In many ways, interacting with Siri is what it might have been like using an early development version of HAl9000, but something is missing. There is no ghost in this machine – it doesn’t feel alive.
We are measuring the quality of AI on a human scale. In fact we measure all intelligence on a human scale; we have no other point of reference. But human intelligence, curiosity, and consciousness are inextricably entwined.
Can an AI be programmed to be curious? Artificial intelligence has proven that it can retrieve answers and perform calculations – but can we program our AI to be creative enough to invent new questions? And if we do, will the AI ask questions that lead to its self-awareness? A sense of being? A will to live? Will the AI suffer the classic existential crisis and start searching for purpose in its existence?
To programmers, making the leap from artificial intelligence to artificial curiosity is a matter of syntax – more code. But to philosophers, the leap from artificial intelligence to artificial curiosity represents the great leap across the Rubicon into the new world of synthetic consciousness.