Cosmic Conversations: Our Mad Dash Towards Super-Intelligent AI
Harnessing Alien Life for Humanity's Benefit with Max Tegmark and Lex Fridman
📽In a recent Lex Fridman podcast (my favorite podcast), physicist Max Tegmark likened AI development to building alien life, a thought-provoking comparison that has profound implications. Tegmark's musings on AI development, human values, and the future of labor provide a fascinating look into how we might harness AI for humanity's benefit while preserving our essence.
42: What is life?
Tegmark believes we hijacked Darwinian biological evolution with AI. In his book, "Life 3.0," he explores the concept of species capable of transcending their biological basis and retaining complexity, even when nature is trying to "mess it up." This hacks these entities into transcending the meat bag — that’s fighting entropy through cellular reproduction. The entities will be able to plug their way into multiple “bodies” and travel the universe at near the speed of light. 📽
Speed Racer: Why are we going so fast?!
AI development has been moving at an unprecedented pace, much like the invention of the airplane. While early inventors focused on mimicking birds, the Wright brothers discovered a simpler way to achieve flight that was less energy efficient. Evolution was constrained because it had to build a flying machine that could:
Repair itself
Fuel/Energy efficiency
In his book, "The Vital Question," Nick Lane touches upon this concept in the section titled "The Threshold of Death." Here, he articulates the idea by stating:
“Imagine you can fly. Gram per gram, you have more than twice the power of a cheetah in full flight, a remarkable combination of strength, aerobic capacity, and lightness. You have no hope of getting airborne if your mitochondria are not practically perfect. Consider the competition for space in your flight muscles. You need myofibrils, of course, the sliding filaments that produce muscle contraction. The more of these you can pack in, the stronger you will be, as the strength of a muscle depends on its cross-sectional area, like a rope. Unlike a rope, however, muscle contraction has to be powered by ATP. To sustain exertion for much more than a minute requires ATP synthesis on the spot. That means you need mitochondria right there in your muscle. They take up space that could otherwise be occupied by more myofibrils.”
Evolution has historically been like a thrifty accountant, trimming the energy budgets of biological systems, such as birds and humans, to save every precious calorie. In contrast, AI is the rebellious teenager with a bottomless stomach, free to devour resources without restraint, while we, the indulgent parents, keep raiding the pantry to satisfy its insatiable appetite. This unfettered freedom has fueled a rapid acceleration in AI advancements.
I discuss the implications and benefits of the renewable energy age in Section 6 of Handbook I: The New Renaissance.
GPT Moloch Babies: They Grow Up So Quickly!
Tegmark's research into deep neural networks and mechanistic interpretability seeks to understand the inner workings of AI "brains." He compares the current stage of AI development to Neanderthals raising Homo sapiens babies, unaware of the potential intelligence explosion as these AI "babies" grow up and outsmart their creators.
Before diving deeper into Tegmark's argument, let's take a brief detour to understand the mythical beast Moloch. Picture an insatiable monster, with fiery eyes and an ever-growing appetite for destruction. Throughout history, Moloch has represented the dark side of human ambition, an allegory for the relentless pursuit of a single goal, often leading to catastrophic consequences.
Now, back to Tegmark's point…
In order to prevent a dystopian future where AI rules over humanity like some sci-fi blockbuster gone awry, he argues that we must instill incentives like compassion and gossip (akin to laws) into AI. These values have helped humanity play a cosmic game of "keep away" with our old nemesis, Moloch. By focusing solely on a single objective function, like capital gains in capitalism, we risk unleashing the beast and letting it run amok, leading to disastrous consequences.
So, in the great battle to save humanity from AI-gone-wild, think of compassion and gossip as our secret weapons, allowing us to sneakily slip past Moloch while keeping our world intact and avoiding the dreaded robot-zombie apocalypse.
Benevolent AI: The Ultimate Potion
Tegmark offers a glimmer of hope, suggesting that we can not only harness AI for humanity's benefit but also preserve our oh-so-precious human essence. By aligning AI with human values, we can conjure up a magical world where wealth abounds, illnesses vanish, and pressing problems crumble before our eyes. AI, armed with the laws of physics, is poised to transform our world into someone’s utopia.
So here comes the million-dollar questions, or perhaps the trillion-dollar ones given the stakes:
Whose values are we aligning AI to?
Are we talking about a wild mix of values thrown into a giant blender, creating an AI smoothie that caters to everyone's taste?
Or perhaps a clash of values, like the M. Night Shyamalan superhero movie Split, where “Patricia” AI takes on different personas and battles itself for dominance?
In a digital tête-à-tête with Max Tegmark, Elon Musk laid down his wager, tweeting:
It seems Musk is confident that if AI should be on a relentless quest for truth, we might just be able to keep the AI apocalypse at bay. Perhaps Elon's grand plan for splurging exorbitant sums on a gasping Twitter is to transform it into a colossal voting system to train his Supreme Justice AI, now christened X.AI. With “truth” and “democracy” as its guiding stars, X.AI — he hopes — will be the AI superhero that leads us to the promised land of harmony and prosperity.
Meanwhile, Eric Drexler, the American engineer who gained fame for his groundbreaking work on molecular nanotechnology and his book "Nanosystems," envisions a future where superintelligent AIs turn our society into a "Paretotopia." 📽 Now that's a fancy term for a world where everyone benefits and no one is left behind – think of it as the utopian garden party where AI plays the gracious host, ensuring everyone gets a slice of the pie, but not evenly. He states his vision here:
“Billionaires might be way up here, competing not to build orbital vehicles but instead starships. And the poor in rural Africa of today merely have orbital space capabilities convenient for families, because they're poor. Nearly everyone much, much better off.”
Tegmark states two critical factors in ensuring alignment:
Compassion
“Gossip”, such as laws that make it illegal to harm people
These are his suggestions of defense mechanisms against destructive forces like Moloch. By incorporating these defenses into AI, we might create a more sustainable objective function that optimizes for diversity of thought and species coexistence.
One thing is clear: Truly aligned AI can solve many of our problems, but we must remain vigilant and proactive in guiding its development and starting the conversation now.
Share your thoughts in the comments. We would love to hear your perspective.
Where is the Pause Button?!
Tegmark believes it is the time to pause, reflect, and consider AI's implications now. By taking a thoughtful approach to AI development, we can harness the power of this alien life form for the benefit of humanity and our future. His analogy of Neanderthals raising Homo sapiens babies also highlights the importance of being cautious and thoughtful about AI development. He believes we are currently moving too quickly, and that competitive economics is pushing us further down a dangerous Moloch path.
Drawing an analogy from the labor movement's success during WWII, Tegmark notes the need to reevaluate our approach to AI as digital and physical autonomy grows. The question remains:
How will our society react to the increasing automation of labor?
Max seems to think that we can indeed enjoy the best of both worlds. By building and carefully aligning AI with humanity's interests, we can avoid creating an AI behemoth that dominates us all. Instead, we can rekindle our affection for being human, cherishing our communities, and nurturing a world where care and compassion take center stage.
Now that's a future worth raising a toast to!
Let's be Prosperous in the Mind
In conclusion, I believe Max's most significant declaration was,
"There is no more guaranteed way to fail than to convince yourself that it's impossible… In military history, the first thing you learn is that's how you do psychological warfare -- you persuade the other side that it is hopeless."
The takeaway? …
Embrace the art of being an optimistic realist. It's the perfect recipe for success in our quest to harness the power of our life’s journey and AI for the greater good.
As you navigate the vast expanse of the Tech Galaxy, remember this guiding star:
"We can only see a short distance ahead, but we can see plenty there that needs to be done."
Alan Turing
I appreciate how your newsletter keeps me up to date on the AI conversation happening with the key leaders in the space. And the way you lay out the environment is like a roadmap of the arguments for and against.
Happy to be a Founding Member - well worth the value you provide.