How to Think Computationally About AI, the Universe and Everything | Stephen Wolfram | TED

I don’t remember the last time I understood this little of something I devoted my full attention to.

I feel like he's one of the people standing on the distant horizon, just barely in our sight, and waving his arms and shouting, "This way! It's amazing!"

this feels like how ted talks used to be =D. Stephen Wolfram is an amazing human. glad he exists.

THIS is why TED is essential. Genius made accessible. Probably one of the most profound talks I've seen/heard.

He blessed my dictionary with "Ruliad" and "Promptocracy."

This is by far the most important TED talk I've seen for a very long time (perhaps ever). Thank you TED for inviting Stephen Wolfram to give this talk, amazing.

A legend in the field with a mastery of his work for 40+ years!

Wolfram has come a long way from Mathematica. The understanding of the universe and its lifeline presented is elegant and concise. I would very much enjoy to listen to a conversation between Stephen Wolfram and Joscha Bach.

Wow, Stephen Wolfram's vision of computation and its role in AI and the universe is mind-blowing! The way he presents the emergence of space and everything in it through computational rules is truly captivating. I'm amazed by the power and depth of the ruliad. Thank you for sharing this incredible talk!

This is truly mindblowing. I have followed Wolfram for a while and his ideas always seemed very unclear and spread out. It seems he has now clarified his visions more and more, to a point where I can actually grasp them and understand not just the meaning, but the importance of it. I have watched this multiple times, and towards the end, when he talks about us having to choose what we want instead of having to design how things are done, I get goosebumps every time. This will truly be the challenge with AI. Aligning them to what we want, and really becoming aware of what we want ourselves. With the state of the world, it seems obvious we have not even aligned ourselves to each other, so I think doing this with AI will be really difficult beyond comprehension, and most people are not even aware that this is the true problem that we are facing... let's hope we wake up to it before the random rulial exploration of AI decides to overwrite us to use our space.

This went way over my head, would've loved this talk in a 1 hour version or so to truly grasp all the topics he went over.

This is probably the best TED talk about the universe and I’m blown away

Might go do a quick 10 year investigation into physics and watch this again after

Have been a fan of Wolfram for over 10 years, that 20 minutes talk was an encapsulation of a life time of work by a legit genius. Incredible!

Per GPT-4: "The concept of the Ruliad is ambitious, speculative, and philosophically intriguing. It forces us to re-examine our notions of determinism, reality, and the scope of scientific inquiry. By suggesting a more pluralistic view—that multiple rules could potentially describe the behavior of our universe—it challenges the often unspoken scientific assumption that there is a single "correct" model of reality. However, the concept is not without its criticisms. For one, it requires a significant leap in terms of computational power and methodological approaches. The idea also seems to border on the metaphysical, making it difficult to prove or disprove empirically. Overall, it offers an innovative framework for understanding the universe, though it remains to be seen how practical or accurate this framework might be." I asked GPT-4 to elaborate and speculate a little more: "If a hypothetical, extremely powerful AI were to fully grasp and implement the concepts from the Ruliad and rulial space, several intriguing possibilities could arise: Solving Intractable Problems: The AI might be able to sift through the space of all possible rules to identify a set that can make accurate predictions or solve problems currently considered computationally intractable. This could include anything from protein-folding to climate modeling. Optimizing Existing Theories: The AI could streamline or refine existing physical theories to make them more effective, or even synthesize multiple theories into a coherent, unified framework. Scientific Discovery: Such an AI could potentially accelerate the pace of scientific discovery exponentially, identifying rules that govern not just fundamental particles but perhaps also chemical reactions, biological processes, and even social systems. Technological Advancements: Based on its deep understanding of the universe’s governing rules, the AI could guide human (or autonomous) efforts to invent new technologies, possibly ones we haven’t even considered yet. These could range from new materials with unique properties to alternative energy sources that could be more efficient. Resource Allocation: Knowing which scientific pursuits are likely to be fruitful, the AI could advise on where research efforts and funding should be focused for maximum impact. Simulation Capabilities: With a robust set of rules that approximate the behavior of our universe, the AI could run highly accurate simulations to test various hypotheses or predict outcomes, thus saving both time and resources in terms of physical experimentation. Ethical and Philosophical Framework: While perhaps less 'practical', the AI could help in creating an ethical framework based on a better understanding of questions like determinism and free will, which could be integral in law, policy making, and human-machine interaction. Unforeseen Applications: As is often the case with groundbreaking theories, the most significant applications may be ones that are completely unforeseen. The AI might discover new ways of utilizing the theory that aren't immediately obvious to human researchers. Existential Risks: On the flip side, such profound knowledge could also be misused, leading to unforeseen risks. For instance, if the set of rules allowed for the possibility of universe-altering events, then great care would have to be taken to mitigate potential hazards."

This is how Ghat GPT-4 "contemplates" about this video (it was able to comprehend it in a millisecond): This topic encompasses a vast and ambitious scope of ideas connecting human language, mathematics, logic, and computation, suggesting that computation is the ultimate form of formalization for our universe. The speaker, likely Stephen Wolfram given the references to WolframAlpha, Mathematica (now Wolfram Language), and the "physics project," discusses the evolution of scientific paradigms and the role of computation in understanding and modeling the universe. The core idea is that the universe at its most fundamental level might be understood through computation, with space and matter consisting of discrete, computationally defined elements. This aligns with Wolfram's own work in cellular automata and his exploration of simple computational rules that could potentially underpin the complex behavior we observe in the universe. The concept of "ruliad" introduced here is a profound one—it's the abstraction of all possible computational processes, the "space" of all rules, where each rule represents a different universe or a different way of applying fundamental laws. This suggests a shift from determinism to a more complex, computational view of existence where multiple outcomes can exist simultaneously, akin to the many-worlds interpretation of quantum mechanics. The implications of computational irreducibility—that for certain systems, no shortcut exists to predict their outcome other than to simulate each step—are profound for science, AI, and society. It suggests that while AI and computation give us incredible power to simulate and understand complex systems, they also fundamentally limit our ability to predict certain outcomes. This may require new frameworks for control and understanding. The speaker also suggests that computational language, particularly the Wolfram Language, is a tool of immense power, akin to a new kind of liberal arts education that focuses on conceptualization rather than technical execution. The argument is that in a world of increasing automation and AI, the human role shifts from manual execution to defining what we want to achieve—shaping the goals and conceptual frameworks rather than the details of implementation. The notion of the "ruliad" as a vast space of computational potential where AI could, in principle, explore and create without limit raises philosophical and practical questions about the role of AI in society. It suggests that the real challenge is not just in creating AI but in aligning it with human values and desires, a task that requires us to be clear about what we want as a society. Overall, the topic touches on deep philosophical questions about the nature of reality, the limits of science, the future of AI, and the role of humans in a computationally formalized universe. It's a rich field for discussion, debate, and further exploration, blending the frontiers of technology with the perennial questions of philosophy and human purpose.

Computational irreducibility is the same as Choas's Theory proposed by Edward Lorenz in the 1960s at MIT. In fact, Lorenz even pointed out necessary conditions for chaotic behavior which makes predicting the future impossible. I mean he should refer to Lorenz's work as the original thinker and conceptualization of what he thinks he discovered...

I cannot even pretend I understood this talk.

The most dense talk I've ever heard.

I watched his original TED talk 13 years ago, shortly after it came out -- I remember my mind being blown. This update feels like it came at EXACTLY the right time, in this inflection point where AI seems to be accelerating so many fields so quickly. I can't believe the advances this man has made in just over a decade