It’s time to learn more about Alexis Maugat 👌

Firstly Alexis could you please tell us a little bit more about your role at 52 let’s say for someone’s just landed from mars?

Alexis: When I joined 52 about 10 years ago, I did not come on my own, but with my little daughter Argine. She is a bridge AI I had been working on for a few years, and ever since I focus on improving her level.

52 was looking for a new IA for it’s bridge application, so that’s the way it all began.

Argine now has a little brother Lancelot, our belote AI. I can tell you raising these two kids is hard work!

Ok but what do you tell your family you do for a living?

Alexis: None of them being bridge players, it was not easy to explain why on earth I would leave my confortable previous job to spend most of my time programming some weird algorithms for a complicated but exciting card game!

But they finally understood that making a living out of your passion somehow looks like heaven!

Talking about Mars what is your professional journey? (Before your position @52).

Alexis: I have a mathematical background (Ecole Centrale of Paris), and strangely enough my first job was to trade interest rate derivatives in a French bank!

After 10 years between Paris and London, practicing a rewarding but stressful activity, I needed to take a break, with no real idea of what I was going to do next.

As it happened, my favorite hobby was to play bridge in competition (see, when I told you I needed to relax !), and I had learned at school the basics of computer programming.

My future now appeared a little clearer to me, I was going to build a bridge robot.

If I had known at that time the amount of work this meant, I would have probably done something else, but once the process is engaged there is almost no turn around!

We have now reached a point where Argine has a very decent level, but improvement that can be done could keep me busy for the next 20 years!

Concretely what is your day to day?

Alexis: Once the bridge robot has reached a “playable” level (ie most player enjoy themselves playing with Argine as a partner), the tough job begins : fixing sharper issues every day with a robot that plays around 1 million boards per day.

So my day to day job is to improve Argine’s level, with Jérôme Rombaut’s help and consumer’s feedbacks.

Bridge is evolving every day, so Argine must keep up to date with new bidding conventions, and a general more and more agressive style.

Back testing is of course crucial, when you correct a bid that was not accurate you have to make sure you do not deteriorate something else .

What did you learn from your first day to today working @52?

Alexis: Humility, just when you think your robot finally has a good level, you realise a lot of things could still be improved!

Having players from almost all over the world, I had to learn and understand many different bidding systems (Polish, Chinese, Scandinavian …) quite some work when for example you have the basic condensed Chinese system that is more than 50 pages long…

You are working with our Chief Bridge officer « Jerome Rombaut » : How do you split your work?

Alexis : Jérôme deals with many issues concerning bridge, but as far as Argine is concerned he gathers all consumer’s feedback, sorts out the relevant ones, and sends them to me to see if it is possible to improve Argine’s bidding or card play.

We then have close discussions to find out the best way to deal with the different issues

Why this name : Argine? (Better than Hal anyway).

Alexis: In french playing cards Argine is the name of the Queen of clubs, and it also is an anagram for Regina (= Queen)

Argine also sounded a bit like Engine, which fitted quite well for a robot!

We hear about a funny feedback from consumers vs IA could you please tell us the story?

Alexis: Bridge is a very irritating game, and when you play with a human partner, arguments and conflicts are commonplaces. When you play facing a robot these disagreements are just the same!

It is very convenient to charge your partner for all the bad boards you had, especially when he cannot retort!

So yes, we have a lot of feedbacks saying Argine really should take bridge lessons, but when you analyse the sequence in detail you often realise the main mistake was done by the human player!

What are your thoughts regarding IA in the gaming industry?

Alexis: Our idea about Argine is not to have an IA that could beat world class bridge players (luckily because we are currently far from it!), but mainly to have players enjoying the game with her.

Therefore Argine must from time to time react like a human, and even if it is sometimes less efficient, consumers surely like it better.

We want to see Argine more like a partner than an opponent, our consumers mainly want to compare themselves with other humans, and they want Argine to play just the same for everybody, so that a good or a bad result will be their own responsibility.

Let’s be serious : Are robots and/or IA will replace humans one day?

Let’s be serious : Are robots and/or IA will replace humans one day?

Alexis: Contrary to Chess or Go where IA now are far superior to humans, actual bridge robots cannot compete with best human players.

One reason for this is that in bridge you don’t see all the four hands, you only have incomplete information when you need to bid or play a card, which toughens the task considerably compared to other game where all the information is available on the board

An other reason concerns reliability of biddings or signals, human players do not always tell the truth at bridge, it is their choice and robots are not always very comfortable with this.

Before an IA can replace top players, it will have to learn to cope with multiple human “behaviors”.

For the time being robots are quite good bridge partners, never complaining and ready to play at all time, so let’s enjoy the game!



Comments are closed