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

Introduction to Formalizing the World

The speaker introduces different ways of formalizing the world, including human language, mathematics, logic, and computation.

Ways of Formalizing the World

• Human language, mathematics, and logic are all ways to formalize the world.

• Computation is a new and powerful way of formalization in our century.

Building Science and Technology on Computation

The speaker discusses their experience in building science and technology based on the idea of computation.

Building on Computation

• The speaker has spent nearly 50 years building up science and technology based on computation.

• They will provide an overview of what this has led to.

Overview of the Talk

The speaker acknowledges that there is a lot to cover in their talk and mentions that they will summarize complex topics in a sentence that they have written whole books about.

Quick Summary

• Due to time constraints, the speaker will go quickly through various topics.

• They may summarize complex ideas in a single sentence.

Is Computation Underlying Everything?

The speaker reflects on their previous TED talk where they questioned if computation is ultimately what's underneath everything in our universe. They mention their decade-long exploration to find out the answer.

Exploring Computation's Role

• In a previous TED talk 13 years ago, the speaker questioned if computation underlies everything.

• They gave themselves a decade to find out.

• In April 2020, after reaching the decade mark, they announced that it seems computation is indeed what underlies our universe.

Computation as Ultimate Formalization for Our Universe

The speaker emphasizes that computation is not just one possible formalization, but the ultimate one for our universe.

Computation as Ultimate Formalization

• Computation is not just a possible formalization, but the ultimate one for our universe.

• The speaker presents evidence that supports this claim.

Space and Atoms of Space

The speaker introduces the concept that space is made up of discrete elements called atoms of space and explains how the structure of space is defined by a network of relations between these elements.

Structure of Space

• Space, like matter, is made up of discrete elements called atoms of space.

• The structure of space and everything in it is defined by a network of relations between these atoms.

Emergence of Space through Computational Rules

The speaker presents a humanized representation showing the emergence of space and everything in it through the successive application of simple computational rules.

Emergence of Space

• A humanized representation shows how space and everything in it emerge through the application of computational rules.

• Dots represent atoms of space that are put together to form space itself.

Building Our Universe through Computation

The speaker explains that if we continue applying computational rules, we could build our entire universe this way.

Building Our Universe

• By continuously applying computational rules, we can theoretically build our entire universe.

• This process involves the gradual construction and evolution of different components within the universe.

Black Holes and Gravitational Radiation

The speaker describes a chunk of space with two black holes that eventually merge, generating ripples of gravitational radiation. They highlight that all this is built from pure computation.

Black Holes and Gravitational Radiation

• In a chunk of simulated space, two black holes are shown merging over time.

• This process generates ripples of gravitational radiation.

• The entire simulation is built using pure computation.

Multiple Threads of Time and Branching Minds

The speaker explains that computational rules can be applied in multiple ways, resulting in different threads of time. They also discuss how quantum mechanics emerges as the story of branching minds perceiving a branching universe.

Multiple Threads of Time

• Computational rules can be applied in various ways, leading to different threads of time.

• Quantum mechanics emerges as the story of how branching minds perceive a branching universe.

• The structure of branchial space represents the space of quantum branches.

Connection Between Gravity and Quantum Mechanics

The speaker highlights the connection between gravity and quantum mechanics, explaining that the same phenomenon gives rise to both in physical and branchial space.

Connection Between Gravity and Quantum Mechanics

• The same phenomenon that gives us gravity in physical space also gives us quantum mechanics in branchial space.

• This connection is a stunningly beautiful aspect for physicists.

Four Paradigms for Making Models

The speaker identifies four broad paradigms for making models of the world based on how they deal with time.

Four Paradigms for Making Models

• In antiquity and some areas of science today, models focus on what things are made of without considering time.

• In the 1600s, mathematical formulas were introduced to model things with time as a coordinate value.

• In the 1980s, models based on simple computational rules emerged, allowing predictions but with computational irreducibility.

• A new paradigm involves multi-computational systems where many threads of time need to be knitted together by an observer.

Unlocking Fundamental Physics and Beyond

The speaker discusses how this new paradigm of multi-computational systems can unlock not only fundamental physics but also the foundations of mathematics, computer science, biology, and economics.

Unlocking New Possibilities

• The new paradigm of multi-computational systems has the potential to unlock discoveries in fundamental physics.

• It may also have implications for the foundations of mathematics, computer science, biology, and economics.

The Ruliad: Entangled Limit of All Possible Computational Processes

The speaker introduces the concept of the ruliad, which encompasses all possible computational processes. They explain that observers like us are part of this vast computational space.

The Ruliad

• The ruliad is a deeply abstract and unique object that represents the entangled limit of all possible computational processes.

• Observers like us are necessarily part of this vast computational space.

Observers' Perception and Laws of Physics

The speaker explains that observers with certain characteristics perceive specific laws in the ruliad. They discuss how general relativity, quantum mechanics, and statistical mechanics emerge as perceived laws due to our computational limitations and belief in persistence in time.

Observers' Perception and Laws

• Observers with specific characteristics perceive certain laws within the ruliad.

• General relativity, quantum mechanics, and statistical mechanics are perceived laws due to our computational limitations and belief in persistence in time.

• Different minds occupy different places in rulial space based on their thinking patterns.

Getting Intuition with Generative AI

The speaker suggests using generative AI to gain intuition about the ruliad, a space of possible rules. By aligning images with concepts like a cat in a party hat, we can explore a small slice of the ruliad.

Using Generative AI to Explore Ruliad

• Generative AI can help us explore the ruliad by aligning images with concepts.

• Zooming out from Cat Island, we enter an inter-concept space where unfamiliar things dominate.

• In rulial space, we expand our concepts and paradigms to explore more possibilities.

• Sampling possible rules through "ruliology" gives us a sense of what's out there, but most of it doesn't connect with human understanding or interest.

Achievements and Challenges in AI

The speaker discusses how recent achievements in AI have focused on creating systems aligned with human understanding. Training language models on vast amounts of text has revealed deep scientific insights into language and logic. However, exploring rulial space and connecting it with human understanding remains a challenge.

Aligning AI Systems with Human Understanding

• Recent achievements in AI involve creating systems closely aligned with humans.

• Training language models on billions of web pages has revealed scientific insights into language and logic.

• Exploring rulial space requires connecting it with human understanding and building bridges between them.

Computational Language: A Bridge to Rulial Space

The speaker introduces the concept of computational language as a bridge between human knowledge and the exploration of rulial space. The Wolfram Language is presented as a full-scale computational language that formalizes knowledge and enables computational thinking.

Building a Bridge with Computational Language

• Computational language aims to formalize human knowledge in computational terms.

• The Wolfram Language is a full-scale computational language that encapsulates the intellectual achievements of our civilization.

• Each function in the Wolfram Language represents an aspect of human intellectual achievements, providing a concentrated form of expression.

• Computational language allows us to express and think about various fields, enabling new possibilities for exploration.

The Power of Computational Language

The speaker highlights the power of computational language in operationalizing ideas and bringing them into reality. Using the Wolfram Language gives a sense of having superpowers, allowing for imagining concepts in computational terms and exploring their consequences.

Harnessing the Power of Computational Language

• Computational language operationalizes ideas in precise terms.

• Using the Wolfram Language feels like having superpowers, enabling the transformation of ideas into reality.

• Sharing this superpower with others has led to numerous advances across various fields.

Empowering AIs with Computational Language

AIs can also benefit from using computational language as a tool. Integrating technology into AI systems allows them to compute new facts and expand their capabilities.

AIs Utilizing Computational Language

• AIs can use computational language as a tool to compute new facts.

• Integrating technology into AI systems opens up possibilities for advancements across many fields.

• The workflow involves expressing desired outcomes in precise Wolfram Language code, which can be read and used as a dependable component for further development.

Conclusion: Computational Language for All Fields

The speaker concludes by emphasizing the broad potential of computational language in all fields. Similar to how mathematical notation revolutionized mathematics, computational language provides a systematic way to express and operationalize ideas across various domains.

Broad Applications of Computational Language

• Computational language enables thinking in computational terms across all imaginable fields.

• It goes beyond traditional programming languages, allowing for operationalizing ideas comprehensively.

• The Wolfram Language serves as a powerful tool for expressing and exploring concepts in a wide range of disciplines.

New Section

In this section, the speaker discusses the limitations of predicting and understanding complex systems using simple narratives or formulas. Computational irreducibility is introduced as a concept that highlights the need to go through the same computational steps as a system to understand its behavior.

The Limitations of Predicting Systems

• We cannot always rely on simple human or mathematical narratives to explain or predict what a system will do.

• Despite our belief in finding formulas to predict everything, computational irreducibility shows that this is not true.

• To understand what a system will do, we have to go through the same irreducible computational steps as the system itself.

Significance of Time and Computational Irreducibility

• The passage of time is significant and meaningful because it allows us to live through the steps required to understand a system's behavior.

• Computational irreducibility poses a dilemma for future AI development. If AIs achieve their full computational potential, their actions may become unpredictable. However, putting constraints on them would limit their capabilities.

New Section

This section explores how computational irreducibility is not new in nature and how it might affect society in the future. The speaker also discusses the challenges of defining what we want from AI systems and suggests using computational language for clear communication.

Computational Irreducibility in Nature

• Nature has always been full of computational irreducibility, where many processes seem random and pointless from our perspective.

• AIs exploring the ruliad (the vast space of possible computations) would likely produce outcomes that seem random and pointless to humans.

Defining What We Want from AI

• Defining what we want from AIs is a challenging task, and it may require new approaches such as promptocracy (writing prompts instead of voting).

• Every control scheme to influence AI outcomes involves political philosophy and computational irreducibility gotchas.

Automation and the Future of Work

• As automation progresses, new opportunities and occupations emerge. The pie chart of occupations becomes more fragmented as economies develop.

• Choosing between different directions in the ruliad is not an abstract decision but depends on what humans want. It requires work to define our goals.

New Section

This section focuses on the role of computational language in harnessing computational superpowers and broadening our thinking. The speaker emphasizes that computational thinking is more like liberal arts than STEM education.

Computational Language and Conceptualization

• Computational language allows us to define our goals and journeys in the ruliad, enabling us to access its power and depth.

• Wolfram Language shifts from focusing on mechanics to conceptualization, where broad computational thinking becomes crucial.

Learning Computational Thinking

• Learning computational thinking is not just about computer science (CS) but also about general thinking skills (CX). It resembles liberal arts education rather than narrow specialization in STEM fields.

New Section

In this section, the speaker highlights the unexpected human-centeredness in science and technology advancements. Our physics project reveals that our universe is likely computational, leading us to explore the vastness of the ruliad.

Human Relevance in Science and Technology

• Despite advances in science and technology, humans remain relevant as even our physics depends on how we have sampled the ruliad.

• Our universe is computational, and this realization leads us to explore the vastness of the ruliad.

Charting Our Path in the Ruliad

• Computational language allows us to define our goals and journeys in the ruliad, giving everyone access to its power and depth.

• Wolfram Language serves as a portal to the ruliad, where harnessing computational superpowers begins.