Video: Superintelligence Panel at Beneficial AI 2017 (FLI)
Elon Musk, Demis Hassabis, Nick Bostrom and more answering basic questions with fascinating answers on trajectory of AI
Don’t have a ton of commentary on this hour of video, other than to say it’s probably the best, most diverse and largest collection of experts I’ve seen speak on the trajectory of artificial intelligence in recent memory. FLI should have called this the Avengers panel or something.
I must admit, I over-consume content on the distant future of AI relative to my fluency in its current capabilities and challenges. In some ways, this isn’t surprising given a recent Morning Consult study that showed the public’s strong support of AI regulation despite it’s admittedly limited knowledge of the subject. We can humbly admit we know little, while still having an opinion on the meta-topic if prompted.
While I’m clearly trying to close my knowledge gap, I also want to be informed about the endlessly interesting meta-discussion. If you feel the same way, this panel from early January will fit the bill.
Laugh Out Loud Moment
At around 32:14, Elon Musk begins sharing his POV on human’s output bandwidth issues (this is months before the Neuralink announcement), and goes on for more than four minutes dropping terms like ‘meat sticks’ in the process. It’s a fascinating monologue as usual, but when he finally concludes it at 36:50, he’s greeted with a hilariously awkward silence and brutal segue to a new topic. As the last panelist to answer every question, this happens to him frequently, but this one really takes the cake.
More Resources from Beneficial AI 2017
- For more videos from the conference, you can check out the FLI YouTube page. There are some great talks led by Yan LeCunn, Yoshua Bengio, Nick Bostrom and Jaan Tallinn along with a few others.
- A set of 23 principles were developed by attendees at the conference this year called the ‘Asilomar AI Principles’. The principles touch on research issues, ethics and values and longer-term Issues. Those principles included were agreed upon by at least 90% of the attendees. To date they’ve been co-signed by nearly 1200 AI and Robotics researches.