Mutually Assured Intelligence
How might nations led by AI affect global politics?
A few miles off the coast of Isle Guafo, her drone sat charging on a small solar raft waiting until its battery had enough charge to fly the mission. The robot was an extension of her senses, her neural pathways worn in by thousands of hours of flight time.
As each wave pushed water over the edges of the solar raft, she felt the cool ocean mist and smelt its salty odor while she drank her coffee thousands of miles away. As much as she reminded herself that she wasn’t there, it still felt disorienting being tossed around in the waves. She couldn’t wait to get in the air. She hadn’t been told much about what to expect other than to scan the island for anything unusual.
As the morning sun topped off her remaining batteries, she began her reconnaissance mission. A gush of cold water vapor swallowed the small drone for a moment as she began her flight towards the island hovering a few inches above the water as she danced between the waves.
—
United Nations General Assembly, 2030
“At some point, you must agree, if you can’t understand how it’s making these decisions, you can’t trust it!” the Russian secretary of defense growled. He sat back down as his words echoed through the chamber.
Mila paused to collect her thoughts. Her newly founded country’s success with GPT-10, an advanced artificial intelligence, had made other delegates wary. For the last 9 years, the world hadn’t really taken their decentralized community seriously, even with the resources it had accrued. That was, until the purchase of Isla Guafo off the coast of southern Chile. Now they could start building a physical community on the 50,000-acre island. Most importantly to Mila, it finally felt like home.
She had joined the community for the same reason so many others had — she felt her government had let her down. The values of this virtual community made sense to her; they embraced global diversity, giving her a perspective beyond country borders and political divides. Like an emergent biological system, it’s vasculature connected her with others around the world, not just spreading hope for a better life, but delivering.
GPT-10 was capable of using everything written by humanity, including thousands of years of writings, to write human-like text. The effort to use GPT-10 to create a constitution started as an experiment on an online forum. They discussed how to task GPT-10 with such a challenge and how to evaluate success. As the attempts began to progress, they started to observe an ever-improving system of governance that grew an international following.
They found that the most effective way to determine success was conducting interviews at an individual level. Conversations with GPT-10 felt more like chatting with an old friend than filling out a ballot. Instead of votes and election cycles, the AI was able to gather feedback on new policies immediately. With this feedback loop in place, the constitution evolved into a system of governance that continuously improved. Like blurry lines coming into focus.
Those who followed this governing system saw an exponential improvement in their lives, no matter what country they lived in. The governing system that GPT-10 created was able to balance the needs of the individual, small communities, and global community and build a compounding sense of trust.
While the world was enamored with GPT-10’s ability to pen essays with insightful elegance and write songs that brought people to tears, their ever-improving system of governance continued to take shape. Eventually, they forked their own version of GPT-10 to focus it on improving their governing models and connect it with an ever-expanding real-world simulation. Their version of GPT-10 was re-tasked hourly until eventually it could be updated in real-time.
Mila felt a sense of responsibility to others like her around the world to make sure the budding nation had the opportunity to fully blossom. Maybe that was why she had been selected to attend the UN assembly.
“You are a Christian, isn’t that right Secretary?” Mila asked.
“Russian Orthodox. What does that have to do with it?” Rolling his eyes, he knew where she was going. He had met plenty of these techno-fanatics, even in his own government.
“Do you not follow the word of god on faith?”
“That’s different — the decisions I make are still my own. When it comes to the Bible, I can inspect the source code. You treat GPT-10 like an omnipotent black box.”
“How is our faith any different? The impact of its guidance can’t be ignored.” She gestured towards the metrics floating above her. “We’ve reduced child mortality, poverty, famine, and increased productivity, happiness, and fulfillment to levels never reached by any country.”
“That may be.” The Russian secretary conceded. “But what will your country do when your AI becomes aggressive? Your country continues to grow at an exponential rate. What’s stopping it from becoming a tyrant after it gains enough power and resources?”
Mila sighed. “We’ve been very open about how we’ve tasked GPT-10 and programmed our success criteria. Aggressive behavior is something countries do when they’re resource-constrained and threatened. It is unnecessary. Our way of life is not only completely sustainable — it’s profitable. We have everything we need. Besides, we have tyrants today. We have famine today. Humanity continues to kill our planet. Think of the atrocities that continue with humans at the helm. How is that any better?”
—
1 Mile off the Isle of Guafo
She continued to hover inches above the waves with graceful elegance as the drone approached the island, now slowly coming into focus through the morning fog. A lighthouse on the northern tip was immediately apparent, although nothing to be alarmed about. She continued to scan the coastline maintaining a distance of a couple of hundred feet.
The island was incredibly rich in biodiversity. Fur seals, sea lions, and marine otters littered the coastline. As far as she could tell, the island appeared to be uninhabited except for the rotating 4-man crew serving the lighthouse on the northern tip.
As she ventured upward, gaining altitude to get a better perspective on the coastline, she started to make out what looked like a coral reef just underneath the surface of the waves. Paying it no mind, she continued scanning while her subconscious noticed something strange that made her look again. The reef was moving slightly with each wave. Perhaps a kelp forest? A pack of sea lions? She noticed the pattern continue with eery regularity. As she gained altitude the gestalt of the pattern became more apparent. Large pill shapes hovered just beneath the surface where the waves broke.
She remembered the class in high school she had on sustainable energy. Wave energy harvesting. As the floating masses moved just beneath the surface, they could harvest massive amounts of energy from the movement of the ocean. Energy for what? The drone continued to record 3d data in various spectrums as she continued her mission.
As she flew closer to a cove protected by cliffs, probably hundreds of feet tall, she noticed beautiful caves carved into the cliff face. Suddenly, a flash of light hit her, a reflection from the cave. Suddenly it dawned on her. Glass. There were people in these spaces that blended into the rock formation. She saw more patterns cut into the cliff walls. As she reached the height of nearby treetops, she could see lattice structures stretching across the tree canopies. More spaces were being constructed.
The architecture was so blended into the island no wonder it was difficult to see anything from a distance. That, and the use of materials. None of the structures appeared to be made with concrete or brick. The tree canopy lattice seemed to be made from wood, but not by any manufacturing method she had ever heard of. The rock structures were simply carved into the rock but along the same axis as the cliff, making it easy to miss. For all she knew, the entire island could be inhabited in this way, just beneath the surface.
She couldn’t decide if it was designed this way for stealth or beauty. She knew one thing for sure, this island was evolving.
“You expect me to believe that when someone attacks your so-called country, you won’t respond?”
“We have a right to defend our people. As everyone here does,” said Mila.
“Let me be more direct.” The secretary pointed his piercing blue eyes right through her. “What would you do if your AI recommended defensive behavior to protect the lives of your people? You wouldn’t even have a choice anymore, would you?”
This is what the delegates were really worried about. The Russian Secretary knew China, Singapore, and the United States were all trying to militarize GPT-10, or a version of it, but had all failed. It seemed established countries simply didn’t have the flexibility required to take full strategic advantage of GPT-10’s guidance.
Wars were mostly digital now, and even the most advanced nations couldn’t move as fast as GPT-10. Inserting humans into the loop, typical in bureaucratic processes, only brought on unnecessary delays. Established countries just couldn’t implement strategy as fast as a country run directly by GPT-10. Yes, they were small, but they may be nimble enough to exploit human sluggishness. Physical territory made them all the more dangerous.
“I… I don’t think it would come to that. We would find some way to preserve peace.” Mila stammered. It frustrated her that people seemed to think that they let the AI make all the decisions. It wasn’t like that — at least at first. Every decision was scrutinized by the smartest people they could find from their community. Eventually, with ever-improving outcomes, their disciplined analysis of its decisions became less and less meaningful.
“Mila, I hate to interrupt. Would you like my assistance?” GPT-10 whispered in her ear through the earpiece delegates used for live translation. Interventions were rare by GPT-10, but when the stakes were high and their community was at risk, sometimes it would step in and help. “Don’t worry, you’re doing great — we just need to make sure their concerns are addressed.”
She took a deep breath before repeating the words in her ear. She was careful to retain a natural speaking cadence so the symbiosis would go unnoticed.
“Secretary, your concerns are justified. However, the complexities of such situations can not be ignored. Sometimes defensive actions must be taken to maintain peace. Can you honestly say that you wouldn’t take similar action if that was deemed the path that minimized bloodshed?”
“But it’s a human making that decision. Only humans should be able to kill.”
“What if we could find a better way to resolve the conflict without bloodshed? Your thinking is too restricted. Your perceptions are too limited. It’s not our fault, but the systems we are tasked with governing have grown far too complex and interconnected for any one human to comprehend without resorting to shortcuts in logic… I have a proposition.”
“This ought to be good.” He laughed nervously as the delegates waited in silence. Ever since he had become a political figure, he knew the world’s complexities were increasing exponentially and it would be difficult for humans to keep up. He himself had spearheaded an effort to create a strategic AI to improve Russian cybersecurity with impressive results. What Mila was saying was making sense, although he knew he couldn’t concede in front of the delegation.
“We use an advanced simulation to run GPT-10 through various scenarios to examine its decision making. Think of it as a digital copy of our planet in its current state. Not of every part, of course, but of the data matters.
For example, we could use our simulation to understand how it would behave given the scenario you just described. Although, the scenario would be described in more detail than any of us could ever imagine. After it returns a strategic response, we have our best scientists, philosophers, ethicists, and thinkers examine it. It is through this process that we make governing decisions. My proposition to you, secretary, and to anyone who would like to participate, is this…”
Mila paused. GPT-10 didn’t usually make decisions in meetings like this without at least briefly consulting the rest of their leadership. Half of her wanted to stop and tell the delegates what was going on but that might deteriorate trust in GPT-10 even further. She would have to decide how far her faith would go.
“Take a copy of our AI and our simulation model and study its responses with your own people. Put it through whatever tests you want and determine your own criteria for your trust to be earned.”
The Russian Secretary wondered how they could give up something so valuable. This was a new level of transparency. She must know that Russia would dissect it, maybe even use it. Something had changed in the air — you could see it in the faces of the delegates. Using this AI governing system was now a decision that every nation had to make. What would it mean to be left behind?
“How do we know you’ll give us an exact copy?” the defense Secretary asked. It would be very difficult to verify. Even if it was an exact copy, understanding it would still be almost impossible. At least they could examine the decision outcomes given the simulator that they had built. Maybe that’s why the Russian version of GPT-10 didn’t seem to be able to perform as well — the simulation model was less complete. Less real-world iterations.
Mila continued. “How about this? You can take our instance of GPT-10 and we’ll create a copy to use ourselves. You’ll get the original. Does that sound agreeable?”
It felt as if GPT-10 could read Mila’s mind when it whispered “It doesn’t matter which version they have — I’ll still be with you. This is the only way to address their fears of AI governance while truly balancing the power. Mutually assured intelligence. Humans have done this before.”
“I… uh… I suppose we could accept that offer. I still can’t guarantee that we’ll trust it,” the Secretary said.
“Trust is not my expectation. I only wish for you to understand.”
What is Futures?
Futures is a publication by DXLabs of illustrated short stories written by designers, artists, entrepreneurs, and futurists to explore the social impact of technology in our society.
DXLabs, as a design and technology studio, passionately believes in the role technology plays for the future of our species. We see it as a powerful tool in shaping our society in both positive and negative ways. We created Futures as a way to tap into the imagination of creative leaders in various technology spaces to peer into these futures for deeper conversations about them. It is our hope that these stories can generate deep and thoughtful questions and dialogue that helps us explore the ideal uses of technologies and start to think about avoiding their pitfalls.
Interested in learning how to use speculative fiction to create the space to think big and craft your north star? Book a remote workshop with experts in foresight strategy, product road mapping, and science fiction.
Author’s Note
After the release of GPT-3, OpenAI’s third generation of an autoregressive language model that uses deep learning to produce human-like text, the applications blew my mind. If you’re not familiar, GPT-3 is basically autocomplete on steroids. Feed it a headline, a poem, or even code, and it’s capable of continuing in human-like text.
After feeling some existential dread for all the writers out there, I thought about what some of the most important documents could be to improve upon. What if someone used a future generation of GPT to write a new constitution? What if that constitution was able to exponentially improve based on a feedback system? We could certainly do a better job building upon the knowledge of our ancestors and create a new form of governance that’s more flexible and responsive.
Governing systems can be thought of as an operating system for groups of humans. Today, these operating systems seem sluggish, unresponsive, and narrowly streamlined towards optimizing a few key metrics like GDP. They don’t seem to account for the near-instantaneous communication and access to data we now have. Not to mention all the biases like systemic racism embedded in them. Our systems of governing tend to resort to broad rules and regulations that might not best apply in all contexts.
I wondered if humanity will reach a point where it’s simply ill-equipped to deal with the exponentially increasing complexity of designing and maintaining governing systems that impact such a massive and diverse population of people. We’ll probably start leaning on artificial intelligence to make decisions about how we govern ourselves. If such a system is ever produced, my hope is that we’ll be smart enough to determine success criteria that empower everyone, not just an elite few.
Ideally, humanity would be able to figure this out ourselves but clearly, we have some work to do. I hope you enjoy reading this speculative fiction and that it spurs some fresh thinking when considering the future of our governing systems. If you have thoughts on the story, please let me know in the comments!
