541. The Ingredients That Make Up Human And Artificial Educability With Leslie Valiant UnSILOed With Greg LaBlanc podcast

Content provided by Greg La Blanc. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Greg La Blanc or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://podcastplayer.com/legal.

unSILOed with Greg LaBlanc »
541. The Ingredients That Make Up Human and Artificial Educability with Leslie Valiant

22h ago 42:01

MP3•Episode home

What does it mean to learn something? While many living things have the capacity for learning, humans have taken this ability to unmatched levels. Our ability to learn and apply knowledge sets us apart from most other species, and now we’re passing that ability on to AI.

Leslie Valiant is a professor of computer science and applied mathematics at Harvard University. His latest book, The Importance of Being Educable: A New Theory of Human Uniqueness, explores our ability to take in new information and raises questions about the broader implications of educability and artificial intelligence.

Leslie and Greg discuss the uniqueness of human educability, how that ability differs from artificial intelligence and machine learning, and the future challenges of integrating machine intelligence in human society.

*unSILOed Podcast is produced by University FM.*

Episode Quotes:

What do people miss when they think about intelligence?

02:05: Well, I think the difficulty is that we don't really know what the word "intelligence" is, and we've been using it for more than a century, and we're using it without having any note of what it means. I don't think it's been very useful, for example, in the study of artificial intelligence. So I think the context of IQ tests, I think, arose in the early 1900s in connection with potential definitions of intelligence in terms of people finding correlations between abilities of children to do various subjects at school. And they hypothesized that the children who are good at many subjects had something, and they hypothesized that what they had, this "something," was this intelligence. But that's not a definition of what intelligence is. So they didn't provide specification of how you recognize someone who's intelligent. It's a purely statistical notion.

What is the best way to understand humans?

03:00: To understand what one is doing, one has to have a definition of what one's trying to achieve. And in some sense, the successes of AI have been along those lines. So, machine learning was something which was defined in terms of what you wanted to achieve. So you had examples of things and you wanted to achieve a prediction of newer examples with high confidence, and people managed to implement this, and this became the kind of backburner of AI. So I think, in understanding humans, I think this is the way forward. We should understand what kind of things we're good at, what we do, what our functions are. And saying someone is intelligent is almost like name-calling.

How can we promote educability without also promoting vulnerability?

39:06: We already have these incredible capabilities for absorbing information, processing it, applying it, running with it. And this capability somehow exceeds our ability to evaluate information. So someone gives us some story about what happened on the other side of the world yesterday. We can't rush over to check it out. We either believe it or we don't believe it. So we find it very hard to evaluate, to evaluate everything we hear.

Show Links:

Recommended Resources:

Alan Turing

Guest Profile:

Faculty Profile at Harvard University
Professional Website

His Work:

526 episodes

#Books #Arts #Business #Greg La Blanc