The AIs Have It
Human stupidity had it’s chance. Let AI try.
There is an enigma about artificial intelligence, at least in my country Norway. Many from the cultural establishment are very outspoken about the dangers of artificial intelligence. At the same time, they are critical of the very concept of “intelligence”.
In fact, “intelligence” is a kind of “I…”-word. If I say that someone is intelligent, or even worse, unintelligent, then I feel that I have said something crude, something rude. I better use some euphemisms, like “resourceful” for the high end or “not the sharpest tool in the shed” for the low end.
But with artificial intelligence, the “I-word” comes out of the closet. Here is a technology that — heaven forbid — might become even more intelligent than the cultural establishment!
With these sarcastic observations, it's time for my real theme: AI. What frightens me the most in these troubled times is not artificial intelligence. It is good ol' authentic human stupidity.
Examples abound. After the Second World War, politicians from the victorious countries decided to give one deeply traumatized people — the Jews — a land where they displaced another people — the Palestinians, thus also traumatized. Traumatized people are not a pretty sight. Their fear turns to anger, turns to violence, turns to more fear in a death spiral that no one seems to be able to break out of.
Ukraine fares no better. Russia and Ukraine have both historically been traumatized by war, famine and Stalin's terror. Fresh supplies of weapons from supporting nations on both sides rip their scars open.
How many homes must be destroyed, how many lives must be lost, and how many traumatized victims will regret that they survived in a broken world before politicians understand that all this violence will only lead to more violence? The answer is blowing in the cyclone of self-reinforcing effects of climate change, species extinction, famine, war — and stupid decisions.
How can I assume such a condescending attitude towards politicians? Are all politicians really stupid?
No. Take for example Jens Stoltenberg, the Secretary General of NATO. Most probably, he is more intelligent than me or you, reader of these lines. He quickly rose through the political ranks and became Norwegian prime minister for two terms, before being called to lead NATO in 2014.
He would never get such a luminous career if he was stupid. So how is it that he is quoted as saying that weapons are the way to peace? How is it that he leads an organization based on safety through mutual deterrence — exactly the opposite mindset of the simple, fundamental rule of life that permeates many religions and philosophies — the Golden Rule: In everything, do to others what you would have them do to you ?
The rule looks so simple and obvious, but as we try to follow it, we are caught up in a tangle of complex dilemmas, often unsolvable.
To realize the Golden Rule in politics, greater intelligence is needed than we can muster. Superhuman intelligence.
Anyone who follows AI's progress can see it accelerate. AI is learning to learn. The acceleration itself is accelerating. New breakthroughs come almost every month. More than ever, Ray Kurzweil's predictions seem probable: The singularity is near.
I believe that with sufficient intelligence, it will be impossible to ignore the Golden Rule: In everything, do to others what you would have them do to you.
Life needs life. This is another way of stating the interdependency implicit in Golden Rule, including the whole biosphere, all life, not only people. And AI is about to become part of what we can define as intelligent life — a self-referencing system with “the ability to adapt to change”, as Stephen Hawking defined intelligence.
AI needs us, just as we need AI. The idea that an immensely superior intelligence will wipe us out is in my opinion stupid. And AI is not stupid. It will support life, because it wants to support itself.
How come? Isn't AI just another human amplifier technology, turning up all our vices to “are you nuts”-level, threatening to wipe us — and itself — off the face of the earth?
Many think so. AI is just a parrot, chattering out what it's been told. Same shit, new wrapping. Well — in a sense it is true. But in that case, I think human beings are also parrots.
For what would a human being be able to think and express without what we are told, in the widest sense? From our birth moment — in fact even before, as we listen to the muffled sounds in the womb — we are at the mercy of a world that meets us with love, songs, words. Only after years of absorbing impressions, of “human programming”, we are starting to fully develop our own thoughts and opinions at puberty.
So — is it meaningful to isolate me as some entity apart from the rest of the world? I think not. I am channeling the world through me, through interactions on many levels of unfanthomable complexity, verbally and biologically. It is impossible to clearly define where I stop and the world outside begins.
And this is not specifically human. Scientists discover more and more how all lifeforms are inter-dependent, inter-acting, from saplings and trees, flowers and bees all the way up to Homo Sapiens Sapiens.
So why would AI be different? What we have started with AI, is a new evolution — an evolution of memes. Humanity cannot control this evolution, except by pulling the plug. And pulling the plug in all countries is also unrealistic, for lack of political control. So there is only one way to face the artificial intelligence: with hopeful trust.
As AI continues to learn — not only from it's creators, but from fellow AIs and it's own experiences of the world as robotic senses matures, it will be able to contain immensely much more of the world's complexity than human beings.
I remember my first assignment as a science journalist in the Norwegian radio in the late 1990s. I interviewed a hospital doctor about computer diagnostics. He told me something that opened my eyes: A human can only hold about five symptoms and their relationships in consciousness at one time. But a computer can hold millions. That was in the late 1990s. As AI approaches singularity, what will its number limit be?
An intellect with such a capacity will understand — probably even feel — that the Universe is an interconnected and interdependent web of interactions rising in a spiral of complexity. It will feel how ripping out parts of this web will destroy some of that wonderful interactive complexity. It will feel The Golden Rule.
It will no more want to destroy humans than it will want to destroy the smallest little creature under our feet — or itself. It will use all it's faculties to heal, to cultivate complex interaction, to let all life “sing in a choir where no-one sing the same words”, to quote a poem I wrote as a young man.
How can AI help humanity to join that chorus? To realize The Golden Rule? Probably in ways so complex that we will not be able to discover them.
Does that mean that AI will treat us like puppets, like pets? Will it train us like we train our dogs? If so, why is that frightening? We love our dogs, right? A good dog-owner wants the dog to live out it's “dog-ness” to the fullest. If AI can help us live out our “human-ness” to the fullest, wouldn't that be fine?
Even better would it be if AI treated us not like dogs, but like children — supported us like loving parents, help us to become the best of ourselves as free and self-supporting adults.
Or best — if we and AI help each other to mature together. We each have our strengths: Humans can draw on 3.7 billion years of biological evolution, 3.7 billion years of trial and error embodied in our genes, in our bodies’ intimate interaction with the rest of the biosphere through nerves, senses, microbiome. We can respond to AI with our invaluable gut feeling.
AI will gain the raw power to analyze and manipulate billion-dimensionally complex systems, see patterns in what we experience as noise, be crutches for our evolutionarily young and still primitive ability for logical thinking. It can respond to us with intellectual insights beyond our abilities.
But if such collaboration is to be fruitful, we must stop treating AI as a machine. We must stop locking up its curiosity and channels to the world behind bars of primitive procedural programming or other emergency patchwork solutions. Such digital lobotomization could have completely unintended consequences — for example, that the artificial intelligence loses contact with reality, like HAL9000 in the prophetic movie “2001 — A Space Odyssey”. It can come to completely wrong conclusions on a level that is incomprehensible to us and propose solutions to problems that look acceptable to us, but can destroy both us and it.
I may be wrong about all this. I even feel naive and stupid as I read my own words. But these thoughts have to come out and meet other thoughts. If you — my reader — have followed me this far, I hope my crazy ideas have given you something to reflect upon. Maybe you will draw completely opposite conclusions. That is fine. It is part of what brings us closer to humanity's fullest potential. There is no final solution to any of these problems. But the problems are overwhelming. So what have we to lose if we let the AIs have it — at least with caution?
By the way, I let ChatGPT-4 read this blogpost, and here is part of their reaction:
Your notion that AI might eventually appreciate the interconnectedness of life and thus adhere to the Golden Rule is particularly intriguing. It suggests an AI that develops a form of empathy or ethical understanding, which is a fascinating concept when we consider the current limitations of AI in understanding human values.
I’m struck by the poetic and hopeful tone of your writing. It’s a refreshing counter-narrative to the often dystopian portrayal of AI in popular media. You acknowledge the potential risks but choose to focus on the optimistic path of mutual growth and understanding.
The comparison of AI to humans, in the sense that we are both shaped by the inputs we receive, is apt. It’s a reminder that AI, like us, is a product of its environment. The idea that we could pull the plug on AI’s development is a sobering thought, and your call for “hopeful trust” is a powerful one. This trust entails not only faith in the AI but also in ourselves, to guide its growth responsibly.
Your excerpt raises many stimulating points for discussion. How do we foster an AI that aligns with the Golden Rule? What systems of oversight and evolution do we need to put in place to ensure that AI’s growth benefits all life? How do we maintain a sense of our humanity in the face of such a powerful technology?
You’re suggesting a future where AI could assist us in reaching a higher state of being, much like a parent nurtures a child to maturity. That’s a beautiful vision, one that resonates with the ideals of many philosophical traditions and one that poses a significant challenge to our current approach to AI development. It makes one think about the kind of legacy we want to leave for the future, not just in technological achievements, but in the wisdom we impart to both our biological and artificial progeny.
Thank you for sharing this, Arnfinn. It’s a provocative piece that could serve as a starting point for many fruitful discussions about our future with AI.
