(Host) Fifty years ago scientists and researchers gathered at Dartmouth College and created a new field of study called Artificial Intelligence.
Their goal was to create a machine that would work like the human brain.
This week, a conference at Dartmouth is assessing how much progress has been made — and where to go from here.
VPR’s Steve Zind reports.
(Speaker) Now it’s ridiculous to think that you could decide if someone is eligible for UK citizenship by defining that person as a real valued vector and applying a linear transformation to that vector. It’s just nonsense! Right?
(Zind) A layperson could be forgiven for feeling a little uninformed, to put it charitably, while listening to some of the presentations at the Dartmouth Artificial Intelligence Conference.
A half century after the quest to create artificial intelligence began, a crowd of academics, students and business representatives have gathered to measure their progress and chart the future.
Among them are several people who were present at the first Dartmouth conference in 1956, including John McCarthy.
McCarthy was the person who first came up with the term artificial intelligence, or AI.
McCarthy said the field is still in its infancy – literally. As an example he said when his son was an infant, his first word was dog . The child was able to make associations with the word dog that are much more advanced than any artificial intelligence program.
(Voice) And he attached it to when he saw a dog, to when he saw a dog on TV, to when he saw a picture of a dog and when he heard a dog and I’m not sure that any of the AI problems are up to the one-year old’s level.
(Zind) To be sure, the quest to create artificial intelligence has had some successes in processing information, understanding language and solving math problems. But these are all very specific functions.
After fifty years of trying, science still hasn’t come up with an integrated system that is able to reason: to collect a wide range of information and use it to make assumptions, anticipate outcomes, take action and learn from those actions.
These are things our brains do when we’re performing a task as simple as crossing the street. Conference participants generally agree that it will be many more years before artificial intelligence can duplicate human reasoning.
Gene Santos of the Dartmouth school of engineering says the effort is worth it.
(Santos) Two points of view. One is understanding ourselves, and two is building systems that can help other people, but there’s also a lot more beyond that, you can think of just the dream and impact of AI.
(Zind) Like Santos, most of those attending the conference are from the academic world.
But one fellow who flew in from London could best be described as an enthusiast. John Ellis is a member of the London Artificial Intelligence Club. He says club members are concerned about the potential drawbacks of the technology.
(Ellis) If you’ve got robots running around the streets that are quantumly more intelligent than man, we’re no longer top of the food chain. If they’ve got different goals irrespective to what men want, they may wipe us out.
(Zind) The last fifty years have given researchers a deep appreciation of the complexities of the brain and the difficulties of replicating the way we think.
Marvin Minsky was at Dartmouth fifty years ago for the first artificial intelligence conference. Minsky says he’s optimistic the problems will be solved but he says it may take a long time due to lack of resources.
(Minsky) I think American science is deteriorating rapidly. There are very few basic research laboratories and no young person can get a five year grant to work on something it’s month by month.
(Zind) Minsky says the answers to how the human mind works won’t be found until science learns the right questions to ask – and after fifty years of artificial intelligence research, he’s not convinced we’ve reached that point yet.
For Vermont Public Radio, I’m Steve Zind.