Lecture Notes: The search for artificial intelligence
While many disciplines study the mind, artificial intelligence (AI) studies the mathematical proccesses by which the brain comes up with solutions. "AI looks at the structure of the problem, and the algorithms used to solve that problem," to try to understand how the brain functions, explains Partha Niyogi, assistant professor in computer science, statistics, and the College.

Niyogi teaches the second part of the two-quarter Introduction to Artificial Intelligence sequence. This year's class-composed of ten math, computer science, psychology, and linguistic concentrators-focused on a single problem for the entire quarter: how to build a speech-recognition software program from scratch. Speech recognition, says Niyogi, is how "the brain takes an acoustic signal and develops a series of representations. At the end of that series is a sequence of words." What is particularly mysterious, even to experts, he adds, is that humans are not born knowing how to do this; rather, it is something we learn.

The project-based course combined lecture and discussion, and the students-who must have already completed an introductory computer-science sequence-worked together on building the system. "They're exploring creatively how one solves this problem," Niyogi says, and they were graded on their participation, as well as the success of their program.
Programming successfully is a daunting task, he explains, because computer scientists do not yet fully understand how the brain maps different acoustic signals-spoken in different voices and accents, amid an enormous amount of background noise- to the same word. "The fact is that we don't know how to solve this problem. So in that way, the students are actually doing research."
-S.A.Z.

---