Deep Learning,

Machine Learning ?

I am so far from being an expert in these fields that it is almost a joke to talk about it
Besides, I let the floor to people who are much more competent than me.

For the novice, it’s about understanding how the latest generations of Artificial Intelligence work – it’s absolutely fascinating – since engineers have managed to go beyond ‘learning from the programmer’ and have managed to imagine models that allow AIs to learn by themselves.

There are lots of different notions and ways to approach the subject of machine learning – the two major ways nowadays – it seems to me – are called Machine Learning and Deep Learning – the one that starts to resemble curiously with the functioning of our brains is deep learning

As long as we stick to pizzas, it’s still okay – well …. it seems okay !

But when you get to the challenge: teaching an AI to drive by itself – without telling it what driving is, what a car is, what the risk is, what the speed is, what the braking is, what the environment is – it becomes incredible :

I was already totally flabbergasted by the concepts used – by this principle of natural selection that will allow to select the following subjects to create a kind of fundamental learning – basic.
I hadn’t seen anything yet – because the next step is reinforcement – improvement – which looks at the whole process and not just the end result. Here it is :

As a result, I was ready to tackle learning something much more complex in terms of expected movements – boxing.
Frankly, besides being mind-blowing, this experience is quite funny to watch – but at one point, the AIpuppets become so badass that it’s enough to leave you dumbfounded :

And what finished me off – it was this experience: the AIs have a role – they have to cooperate to win – and they have complete freedom of action.
So, you will see, how each team literally imagines strategies to win – taking into account both the environment, the rules of the “game” – in terms of time and spatial and physical constraints :

The AI that amazes researchers has behaved intelligently – there is no doubt about that. It has “broken” the given framework to use its environment to the maximum. That’s just magic.

What to make of this for Altair?
I’ve just been blown away – right now my ideas are not yet clear, although I have a couple of things on the horizon.
What I do know is that by inventing a concept where AI is welcomed and leveraged to optimize all the experiences on offer, that concept will not only optimize but also surprise the audience.

If you have any doubts about how AIs can surprise us with our reasoning patterns, check out this little joke below.

In reality, it’s impossible to think about the way the “horse” moves/runs – in virtuality, of course it works. It is very strange – but it works.

That means ? it means that letting AIs propose “games” in virtual reality, letting AIs propose games and play in the AA App, it will improve incredibly the field of possibilities – AI is not in our reality – it doesn’t care about shopping at wallmart – when it has acquired intelligence it invents processes for “its” world.
Frankly, the idea is worth exploring – deeply.

Home Page

Featured Image : Legion – famous IA character developed in the Mass Effect video games ( Bioware)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s