Okay.

Welcome everybody on this wonderful sunny day.

I hope this is the lecture you're waiting for.

Because that's the lecture you're going to get.

Let me see.

Is this better?

Okay, so this is the AI2 lecture.

Who of you attended last semester's AI1 lecture?

Okay, that's a sizable portion.

All of the things we did simple last semester, we're going to do difficult this semester.

Okay, where we had static environments, we're going to have dynamic ones.

Where we had fully observable environments, we're going to have partially observable environments.

What else? Where we had deterministic actions, we're going to have unsure actions.

All of those kind of things.

Okay, and when we've done that, we're doing machine learning.

Okay, just the basics of it, because there are lots of specialized machine learning lectures at FAU.

And I'm assuming that you've taken a couple of those.

Okay, so before we start, you remember that I insist that you are engaged and ask questions.

That is still true this semester.

Okay, so please do not be shy. Ask questions.

If you have a question, don't wave like this.

Make sure that I notice.

And I'm very happy that so many of you are here.

We must at least have 15% of everybody who is registered on Stodon for this course.

So to the other 85%, it would be really good for you if you were here as well.

If you have visa problems, you're excused for a couple of weeks.

Otherwise, it's a good idea to come to the lectures.

Okay, are there any questions?

Okay, so you probably know the drill.

We do the intro today, the admin and all the boring stuff,

so that we clear them out of the way for the more interesting stuff that's going to come.

Okay, so if you've asked yourselves, what should I learn in AI too?

Well, really just like the last time, it's going to be kind of a course that gives you a walk through the vegetable garden

of statistical and sub-symbolic AI, and we kind of sniff at every plant,

but we're not actually going to go deep anywhere.

It's more about the holistic big picture of stuff, plus quite a lot of tools and algorithms.

And so really, it's about mathematical or quasi-mathematical.

We're not going to use the full power of math here.

Models for certain things that our biological models of intelligence can actually do.

And of course, the underlying principles of those models.

What are the assumptions? What are the limitations? What's the math behind them?

So just like we did for the easy stuff last semester, we're going to do for the more interesting environments

that we want to study this semester.

Also important, you should have at the end the ability of actually formulating real-world problems in terms of these models

and see which models are adequate. That's an extremely important skill. I'll come back to that.

So really, we're interested in what are the phenomena we want to model,

the phenomena of intelligence, like being able to deal with partially observable environments

and behaving well and successfully in them.

And what are the models like? What are their limitations? Are they even applicable to the problem I'm trying to solve?

And then of course, algorithms. And the ideas behind the algorithms.