Dr. Fei-Fei Li

But if I say the word one plus, it's not like...

But if I say the word one plus, it's not like...

5 typically follows or 2 typically follows, right?

5 typically follows or 2 typically follows, right?

Like, there is actually a deeper rule of 1 plus 2 equals 3.

Like, there is actually a deeper rule of 1 plus 2 equals 3.

Of course, when it has seen enough of that, it should predict 3 for today's language model.

Of course, when it has seen enough of that, it should predict 3 for today's language model.

And actually, it does.

And actually, it does.

This is too simple an example.

This is too simple an example.

But the point is that...

But the point is that...

Math takes a higher level of reasoning than just following statistical patterns, and large language model by and large follows statistical patterns.

Math takes a higher level of reasoning than just following statistical patterns, and large language model by and large follows statistical patterns.

So some of the mathematical reasoning is lacking.

So some of the mathematical reasoning is lacking.

Yeah, this title came about after I finished writing the book.

Yeah, this title came about after I finished writing the book.