Why has AI been taking so long to develop compared with its fictional version?
These are my notes from the SXSWi panel, which asked the question. The chair was Dr Doug Lenat of Austin based www.cyc.com (email@example.com):
“Cycorp was founded in 1994 to research, develop, and commercialize Artificial Intelligence. Cycorp’s vision is to create the world’s first true artificial intelligence, having both common sense and the ability to reason with it.”
Doug Lenat shows a brief history of AI in both science fiction and science fact. He says that he believes the HAL 9000 computer in the 1968 film 2001 Space Odyssey, showed us a way of changing everyday life on Earth. However, he asks, twenty years later, is the Alice chatbot as good as it gets? We continue to underestimate the problem of how much information we are using as humans when we use language.
Lenat takes a shot at predicting the future of our relationship with computers over the next ten years:
2013: crowd sourcing explosion.
2015: computers capable of question answering, semantic search and syntactic search.
2020: cradle to grave mental prothesis – we’ll delegate more and more to our mechanical servants.
The panel comprised of: Professor Peter Stone from University of Texas at Austin department of Computer Science and director of Learning Agents. His research has taken the goal that the movies have adopted ie fully autonomous agents in the real world. He believes we will try to build complete agents on a closed loop ie robots that sense, decide and act. This will drive research forward he believes. He uses the example of Robocup project where robots play football. By 2006 robots are fully autonomous individuals on soccer field and they do team work. He also cites the DARPA Grand Challenge wherein a robot drove across the desert and also examples of robots driving cars on race tracks. He shows a simulation of how robots driving cars could solve traffic flow problems through automation and detection. He concludes that AI can be part of a solution for a good future: I imagine he means compared with the dystopian vision of the Terminator franchise in which machines take control and seek to destroy all human kind.
Next up is Natasha Vita More. A strangely beautiful woman of 60 who yet looks somewhat robotic herself. I can’t say she has definitely had surgery but she was, as well as being the only woman on the panel, the one who looked and sounded most like a visitor from the future. She is currently doing her Phd at the University of Plymouth and her subject is the Transhuman or the Post Human.
Vita More asks: what is human enhancement? What are its media? In future will here be a new criteria for normal? And as for behaviour, will biosynthetic identities be familiar or fearful?
What will happen if a biologically unfixed species develops? What happens if we create multiple bio-syn personas? Will we consider ourselves disabled in the future if we’re not transformed or enhanced by AI?
Doug Lenat rounds up with how a fiction writer might look at the idea of human plus machine from an artistic perspective. A question from the floor asks whether computers can ever make something ‘ beautiful’. Lenat says it is in the eye of the beholder and mentions Harold Cohen’s programme, which has made ‘beautiful artwork’. He says in a century or two people won’t understand what this problem is about, or why we’re concerned with it. A follow up question asks about the possibilities for humour as machines learn to be more like us. Lenat says it is a question of whether a computer can predict what will be seen as funny or beautiful in any context. I think that what is funny or artistic is often something which is out of context. Another layer of complexity of meaning which computers will need to understand.
Lenat concludes by saying that as well as a sense of self awareness and self purpose, we must build a sense of morality into these machines. Whose morality in what context remains to be seen.
The AAAI Presidential Panel on Long Term Futures (Association for the Advancement of Artificial Intelligence) www.aaai.org
Leave a Reply
You must be logged in to post a comment.