A HOT NEW BOMBSHELL ENTERS THE VILLA: COOPERATIVE AI

Put a finger down if you’ve ever decided to play Overcooked with your partner/friend on a chill Saturday evening, thinking it would be a cute bonding experience—teaching you the value of teamwork, just in case you both ever end up as crew on Captain Jack Sparrow's ship.

But what started as a cutesy “let’s make virtual tomato soup” session quickly turns into chaos. Suddenly, you and your partner/friend  are screaming at each other, trying to avoid burning the kitchen down. You both become real-life Mr. Krabs and Plankton, locked in a battle over who gets to control the secret recipe. Your sweet(pun intended), loving partner is now a tomato-chopping tyrant, delivering orders faster than you can keep up—and doing it all without asking for help.

You put your finger down, didn’t you? I knew it.

While Overcooked might just seem like a game, it’s actually a perfect example of how important collaboration and strategy are—because without them, you’ll be stuck in a virtual kitchen nightmare. But when you do work together, things start to click. Before you know it, you’ve got three stars, your relationship survives another day, and maybe, just maybe, you’ve learned to cook (virtually) with your human.

Now, this is kinda what Cooperative AI is all about.

Let’s face it—AI is everywhere. If we’re voting for “Word of the Year,” AI is winning by a landslide (cue applause). From your phone to your work, almost every company has AI integrated into their ecosystem. But here’s the catch: it’s not just enough for AI to be smart; it needs to know how to work well with humans, other AIs, and even machines. 

You’re probably wondering, "What even is Cooperative AI?" Well, let me break it down for you.

Cooperative AI is about teaching AI systems to align better with human goals, make decisions together, and collaborate more effectively—whether with humans, other AIs, or machines. The big idea is to have AI systems that can work with us, not against us, in making decisions, planning strategies, or solving problems. And here’s the kicker: sometimes, it's not just humans and AI working together. You could have AI-AI collaborations, or teams with humans and multiple AIs. It's like assembling an Avengers team, but for decision-making.

 

To get AI on the right track, these systems are trained to work towards a common goal. And every time AI successfully collaborates with others—whether humans or other AIs—it gets a reward. It’s like the Overcooked moment when you finally stop arguing and start working together—suddenly, everything flows, and you get that perfect score.

But why does this matter?

Because the world is full of situations where we need to work together. Climate change? Global conflicts? Even small stuff like who rides shotgun in the car on the way to school. Cooperation is key, and cooperative AI could be a huge part of helping solve these challenges. Whoa, now listen I am not saying it’s a magic solution for every cooperation problem in the world, but it’s definitely something worth exploring.

You know that phrase, "It's just one plastic bottle," said seven billion people? That’s the problem we face in a nutshell—small, individual decisions add up. Now imagine that but with millions of AIs, each making decisions without thinking about the bigger picture. With Cooperative AI, the goal is to ensure that every system—whether it’s humans, AIs, or both—works together to achieve better outcomes for everyone.

So, next time your kids are fighting over who gets the passenger seat and the schedule you created for them isn’t working, think about it. Could AI solve this problem? Maybe!

Congratulations, dear reader! Your reward is a (hopefully) better understanding of Cooperative AI. If this has tickled your fancy, stay tuned for the next article, where we’ll explore how we can actually make AI systems more cooperative.

Resources:Introduction to Cooperative AI