Okay, so...

Yes.

I am charging, so we won't have a dark out this time.

So please, I asked him to remind me, but if he is ever not here, please remind me.

Because I have a course before that and it's early in the morning and I always forget.

And so this laptop kind of survives two lectures.

But if it's also converting the recording of the first in the background,

then we're missing five minutes or something like that.

And you've experienced this.

So it's always good to remind because when I'm in lecture theatre,

the adrenaline is flowing freely and I forget certain things.

Okay, so we're still talking about logic.

And will be for a while, namely inference and so on.

And you've yesterday, Florian Rabbe, did a couple of examples for a logic,

a new logic I call PLNQ, predicate logic without quantifiers, no quantifiers.

And the idea why we, or I want to introduce that logic is because it's equivalent to propositional logic,

what we've been talking about.

But it's also nearer to this idea of structured world representations.

Remember, I tried way back two months ago or one month ago to convince you that we can either have atomic states,

which are so, so simple that we're left with nothing but using search,

or factored world representations where we can do better things, namely constraint satisfaction.

And there's the structured representation of worlds.

There was this example that you want to pass a street,

but there's a truck back up which cannot back up into a driveway because there's a cow.

And if you find this a somewhat constructed example, well it is.

I want to make the case with this example that we can actually do, no matter how crazy it is, we can actually represent it.

Now, how would we do this?

Well, we would do it exactly like you've done it.

We would just basically give ourselves a signature.

We would have something like C for a cow.

We would have T for a truck.

Those are individuals.

We would have D for a driveway, and we would say the cow is in the driveway.

The cow C is in the driveway D.

And we would say if there is a truck, no.

And we can have something like a predicate can back up a truck, which means get out of the way or something like this.

And with all of this, we would be able to describe the situation.

The agent would see the situation, would understand what's going on, and would create this kind of a vocabulary which we can actually express in PLNQ.

Okay?

Now, what would the agent do then?

It would see a situation which is interpreted as this.

It would represent the whole thing in PLNQ.

It would then remember, ah, PLNQ is exactly propositional logic.

So I can use my propositional reasoner to know what my actions do and what the best way out of this is.

Okay?

That's what an agent does.

Remember the Wumpers world, right?

Where we know things about the world.

Here, we've actually written them down as propositional variables.

In PLNQ, we can write them down a little bit more nicely.