Welcome to AIR1 in 2026.

I'm quite happy to see so many comparatively of you here in spite of the snow.

I must say I've had nicer bike rides to the university than today and I'm assuming something

similar applies to you.

So all the better that you're here.

We are doing symbolic AI and that basically means we're designing artificial agents.

That means we're designing software

possibly hardware programs

that can interact with

the world.

They have ways of sensing the environment.

That could be eyes or ears or something like this.

Or just a text interface

whatever you want.

And that could act on the environment.

For instance

like a chess player who moves pieces or think of a robot that travels in

Romania or something like this.

And the main class of agents that we've looked at and are still looking at are goal or model

based agents in a very simple class of environments.

The environments are static

meaning nothing is changed unless the agent changes it by

its actions.

Completely unrealistic.

The only agents that kind of have this live in static environments where only things change

are gods.

It doesn't rain unless the god makes it rain.

We're not building artificial gods.

It's AI, not AG.

So the other thing about the environment is it's fully observable

which again is completely

different to the environment we work in or we live in.

And the environments are deterministic.

If we do an action, it succeeds.

Again, completely oversimplified.

But looking at these kind of agents in this very simple case is a wonderful kind of set

of training exercises to proceed to the more interesting and important environment

which

are of course algorithmically much more interesting

we could say

or much more difficult.

And of course

there are agents to whom the world is fully observable.

And think of the thermostat down there.

All that thermostat

quote unquote

wants to do is keep the temperature constant.

And it really doesn't care if there are students here or not.

So the temperature is the only variable of the environment that this little agent actually

has to observe.