Did Star Wars predict our AI future?

Yet in all this uncertainty, there seems to be consensus over one point: 2025 will be the year of AI Agents. AI systems have become more than digital tools - they are behaving like assistants. If you've ever used ChatGPT, you may have noticed that you interact with it more like you would with a colleague or an intern, and less as a piece of software.
The fact AI can use tools allows you to interact with technology at the level of objectives, rather than task execution. All of this is opening many new opportunities to completely rethink our relationship with technology and core concepts of organisations, division of labour and even our place in the future World.

What is the right way to do this? Should we rely on the monolithic AGI Sam Altman is promising us? Or should we stick to Copilot and ChatGPT? Should we use AI at all, or perhaps only for certain tasks and roles?

Once again, I turned to Hollywood.

Many movies explore the topic of AI from various perspectives, but there is one that approaches it in a distinctive way: Star Wars. It took me a while to think about it, but today, I believe its portrayal of AI presents a vision that resonates with many of the major challenges we encounter with AI today and could offer a glimpse into our AI future.

If you were born before 1990 you can probably skip the next few paragraphs, but for all of you Gen-Y/Z-ers, I'll provide a brief overview of the plot. I'll focus on the original episodes 4-6 only, because, let's face it, the world would have been a better place without episodes 1-3. And while I love episodes 7-9, it would take too long to explain the plot without adding much to the discussion.

Star Wars is a space saga about the epic battle of good versus evil. Technology is advanced enough to provide ability of interstellar travel faster than the speed of light, lightsabers and micro-regenerative medicine. For some reason, however, time is measured in parsecs.
At its core, the universe is Taoist-like. All life generates the Force, an omni-present field which permeates all of the universe and binds it together.
The Force has two inseparable sides: a light side associated with passiveness, compassion, and selflessness and a dark side, associated with aggression and destruction. People who learn to use the Force acquire enhanced consciousness and superhuman abilities such as moving objects with their thoughts. On the light side they are called the Jedi, and on the dark side the Sith.

The Star Wars saga revolves around Luke Skywalker, a moisture farmer from the desert planet Tatooine in a galaxy far far away. There is something special about Luke: even though he doesn't know it, he is the son Darth Vader, a powerful Jedi-turned-Sith who is bent on conquering the galaxy together with his mentor, the galactic emperor.
Through a set of planned circumstances, Luke also becomes a Jedi and goes on to fight his own father and the emperor, to destroy the dark side, bring the balance to the Force and peace to the galaxy.

At this point, if you are wondering, the answer is yes: the complexity of toxic family relations in Star Wars is on par with Steinbeck's "East of Eden"... but this is a topic for another article.

Back to StarWars... AI exists in the Star Wars universe, but not as all powerful, monolithic AI. Instead, specialised robots (droids) work in teams with other galactic beings to achieve common objectives. There are droids whose job is to translate languages and facilitate communication between different species, others who help maintain spaceships and navigate "hyperspace", some that clean, others that serve as medical staff etc.

Luke Skywalker doesn't act alone in his adventure. He gathers a rag-tag team consisting of a smuggler Han Solo, his faithful pilot companion Chewbacca, droids R2D2 and C3P0 and his sister Leia. Skywalker's multi-agent team functions in a way which is more than the mere sum of the parts. Luke needs Han Solo for his super fast spaceship, the Millennium Falcon, and Chewbacca to pilot it.
Meanwhile he also needs C3PO to translate between him, Chewbacca and other non-humans. R2D2 is there to help maintain the ship, navigate hyperspace and at times serve as a glorified USB stick (he performs the critical role of smuggling messages between rebel groups).

The humans in turn protect the droids from danger. You get the point... the objective of destroying the galactic emperor is achieved not by any one individual, but as a hybrid team of AI and humans. The team works without interruptions and behaviour and personalities of each member are dynamic.

This is exactly what people are trying to build nowadays with multi-agent AI systems!

As crazy as it may sound, the way AI is portrayed in Star Wars may well be the way we end up adopting AI into our society and organisations. It is interesting that the first Star Wars movie came out in 1977, 4 years after the ground-braking paper of Hewit et. al. on the actor model which laid down the core concepts for building and operating multi-agent systems.

Why are multi-agents so interesting? 2025 may be the year of the AI Agents, but the conversation is, at least for the moment, revolving mostly around the interaction of a single human with a single agent. Naturally, the next step will be to ask ourselves: can we build multi-agent teams, where multiple humans and multiple AI agents work together to achieve a common objective? Some of you may immediately declare: "of course we can!", but this is not so trivial. Building a 1 human - 1 AI agent system is one thing, but once you have more than 2 members in the team you have to be able to answer a host of new questions:

• How do I deal with the fact that humans and AI function on different time scales without stopping the team?
• How does this system scale to a large team?
• How should the communication between agents be orchestrated?
• How do I avoid acting on partial information or information cross-over?
• How do I deal with the fact that the behaviour of agents can evolve dynamically?

Yet despite the challenges, multi-agent systems may help solve many of the issues we currently have with our AI future, which makes them a very attractive approach to AI development.

Multi-agents are by definition teams, and hence more easily inclusive and fair than, say, the full automation approach to AI. They also allow much more control over governance and accountability compared to monolithic AI systems. With multi-agents, we get do decide how we divide the roles and labour as well as the rules of behaviour and resource access, as we do already! The whole thing has a much more human feel to it compared to monolithic AI.

It may even be that the colourful dynamics of the rag-tag team of Jedi, smugglers and droids which makes Star Wars so relatable ends up being the main quality which makes multi-agents easy to adopt by the society - it is, at its core, a human centric approach.

Monolithic AI models such as GPT-4 have impressive capabilities, which they draw from the fact that their information universe is that of the internet. But this is also the limitation of monolithic AI. We are exhausting the useful data we can extract to improve these models, and synthetic data generated by the same models by definition can not bring new information into the system (keep in mind that information is not the same thing as new data).
Multi-agents instead rely on collaborative intelligence. This is very important, because it allows new information to enter the team through human interaction as well as to address core issues such as AI model bias, as you don't have to rely on a single AI model.

Multi-agents are a lot more sustainable compared to monolithic AI. Agents can be designed in a way that consumes resources only when needed. In Star Wars words, you don't need to fire up the Millennium Falcon each time Luke Skywalker wants to turns on his lightsaber. The fact that resources can be used and allocated in an atomic manner could help alleviate, at least in part, the troubling trends of energy consumption due to AI use.

Finally, multi-agent systems are resilient. If one team member crashes, the team can continue. If C3PO from StarWars was damaged, you could replace it with another protocol droid and the team continues. This would not be the case if one relied on monolithic AI. Once a monolithic model is down, the whole thing is down.
Imagine Star Wars where Luke Skywalker fought the emperor, navigated the galaxy, translated between different galactic languages, flew the spaceship, and occupied all other team roles at once... and then Darth Vader comes by and cuts of his hand, moments before he reveals that he is, in fact, Luke's father.
This is in fact what happens in episode V... I'm telling you, Steinbeck level of f-ed up family relations. I could hardly imagine a satisfying ending to the saga in this scenario.

The more I try to imagine a sustainable future with AI, the more I am convinced that multi-agents is the way to go forward with AI. All the pieces start to click into place once you start thinking about AI as a member of a team, instead of a tool or a vehicle for process automation. I am excited about this because we are at the right moment to make decisions on how we want to live with AI in the future... and yes, it is still our choice.

Nothing is inevitable, but to me at least, it appears that a great way forward with AI has shown itself! It's up to us to decide: light side or the dark side.

Enjoyed this insight?
Share it to your network.