Version 2.01 no time to cry
lecture: The quest for artificial general intelligence
beyond the Turing test
A whirlwind tour of the past, present, and future for artificial general intelligence (also known as strong AI). This talk will begin with Alan Turing's original test. This is not the same as *the* Turing test everyone has heard of. We will look at the current state of the art covering such topics as:
* How old does a human have to be to beat the current state of the art AI in answering science questions?
* What are some other ways we can test AI?
* How do we teach computers language?
* How well can AI follow a written story?
* How well can AI recognize what's in an image?
* What basic things are easy for humans to do but very difficult for AI?
* What are the common objections to the advancement of AI?
* What does the future hold?
As with anything that pushes the boundaries of what's possible, we'll also take a look at some hilarious failures with chat bots and image scene recognition.
No prior knowledge about artificial intelligence is required for this talk.
This talk will begin with a definition and a quick history of AI and some famous failed predictions. I'll cover the curious origin of Alan Turing's formulation of his test.
The talk will cover general approaches in areas like how we teach AI to understand science, language, and images. I will address persistent problems, the current state of the art, and the formalization of tests preceding and succeeding the Turing test.
The talk will conclude with the possible future and what happens after the Turing test is passed: including a discussion of whether recent caution about unconstrained AI is really warranted.
Start time: 11:15
Room: Project 2501