The theme of Humanity Plus’s third session was “Redefining Intelligence: Artificial Intelligence, Intelligence Enhancement, and Substrate Independent Minds.” Speakers talked about Artificial Intelligence, Artificial General Intelligence, how to create brains, and everything in between.
Far and away, this was the best session of the event. Not only were the speakers well-prepared and good presenters, but the topics were generally the most engaging and well-researched.
One of the most prevalent topics throughout this session was the idea of defining intelligence. Randal Koene, who spoke first, said, “There’s no single algorithm for general intelligence of the brain.” Rather, Koene said, it’s determined by a variety of factors and is the result of a long and storied evolutionary history. Suzanne Gildert, the second speaker of the evening, agreed that it wouldn’t be easy to create a general purpose brain system. However, she had a very unusual definition of intelligence that was very data-oriented, even going to far as to say that a photon is an intelligent system, a claim that didn’t go over very well with the audience. Geordie Rose looked at intelligence from a more neuroscientific perspective, whereas Paul Rosenbloom noted that an intelligent machine would have self reflection. All in all, nobody seemed to have a concrete definition of intelligence, but all took the time to define it at the beginning of their talk.
Two talks in particular stood out in this series. The first was Alex Peake, from Primer Labs. Primer Labs is working on the OpenPrimer project, which aims to “develop a free game and web platform that makes any game a lifelong learning game and embeds game learning into any web content.” Peake was an incredible presenter. One of the themes I discussed in an earlier blog post was the necessity of making science accessible to the general public, and this was a perfect example of doing just that; in the first five minutes of the talk, he referenced Neal Stephenson, Earl Grey tea, Brave New World, and Jean-Luc Picard – complete with the accent. However, Peake also discussed more philosophical matters, such as “What happens when we make machines that can engineer humans?” and “Will we love our kids – and will they love us – if they relate to artificial intelligence better than they do to us?”
The second talk that really stood out was Paul Rosenbloom‘s. Hailing from the University of Southern California, his claim to fame at the conference was a very impressive model of Artificial Intelligence, complete with videos. He also brought definitions to the forefront of his talk. Rosenbloom said, “A virtual human is a graphical human body with a simulation of a human mind inside of it.”
Overall, a really excellent session. My main takeaway message was that we don’t yet have a solid definition of intelligence. Right now, it seems like the entire scientific community is approaching it haphazardly, from a number of directions. The rest of us are, at the moment, content with the “I’ll know it when I see it” approach.