What are some key functions or capabilities of Wolfram Language?

Wolfram Language has around 6,000 primitive functions covering a vast range of domains. Examples include 'ImageIdentify' for object recognition in images and 'GeoNearest' for finding locations like volcanoes, demonstrating its ability to connect to real-world data and perform complex computations.

How does Wolfram Language relate to AI and machine learning?

Wolfram Language integrates symbolic knowledge with modern machine learning techniques. While machine learning can learn from data, Wolfram Language focuses on encoding existing knowledge computationally, creating a robust foundation that machine learning can then explore and enhance.

Why isn't Wolfram Language more widely popular?

The adoption of complex ideas like Wolfram Language can be slow, influenced by factors like egos, personalities, and social network dynamics. While highly used by top researchers, its 'trickle-down' to broader adoption has been slower, partly due to the company's focus on product development over aggressive commercialization.

What is Wolfram Alpha and how is its knowledge base built?

Wolfram Alpha is a question-answering system that uses Wolfram Language and a vast, continuously updated knowledge base. It was built by systematically ingesting information from reference libraries and consulting experts in various fields, aiming to provide computable knowledge on a global scale.

What are computational contracts and how might they change legal agreements?

Computational contracts represent all conditions of a contract in a computable form, allowing for automated execution and streamlined processes, rather than relying on traditional legalese. This idea, envisioned centuries ago by Leibniz, is now becoming realistic with advanced computational languages.

What is symbolic discourse language and its relation to the Turing Test?

Symbolic discourse language aims to represent everything in the world in a precise, computable, symbolic form, extending beyond contracts to common sense notions. While the Turing Test focuses on mimicking human conversation, Wolfram Alpha's strength lies in its deep knowledge base, arguably addressing a different, perhaps more fundamental, aspect of intelligence.

Key Moments

What is Wolfram Language? (Stephen Wolfram) | AI Podcast Clips

Lex Fridman

Science & Technology4 min read41 min video

Apr 21, 2020|75,121 views|1,735|151

stephen wolfram artificial intelligence ai ai podcast artificial intelligence podcast lex clips lex fridman lex podcast lex mit lex ai mit ai ai podcast clips

Save to Pod

Key Moments

TL;DR

Wolfram Language is a high-level symbolic language aiming to represent the world computationally, powering tools like Mathematica and Wolfram Alpha.

Key Insights

Wolfram Language is a symbolic, high-level programming language designed to represent concepts and things in the world rather than just computer operations.

It powers tools like Mathematica (since 1988) and Wolfram Alpha, aiming to create a comprehensive computational knowledge base.

The language includes a vast array of built-in functions (6,000+) covering diverse areas like image identification, geographical data, and user interfaces.

Wolfram Alpha's knowledge base was built by ingesting vast amounts of information, leveraging expert input and a methodology of implementing areas incrementally.

There's an ongoing effort to bridge symbolic computation with machine learning, creating a hybrid approach for deeper understanding and application.

The future vision includes 'symbolic discourse language' to represent complex human interactions, ethics, and legal contracts computationally.

THE CORE CONCEPT OF WOLFRAM LANGUAGE

Wolfram Language is fundamentally different from traditional programming languages. Instead of focusing on what computers inherently do, it's designed to work with concepts and things that exist in the world or can be imagined. It boasts a symbolic nature, meaning it treats elements like 'X' not as an undefined variable but as a potential representation of a real-world concept, such as the city of Boston or the trajectory of a spacecraft. This allows for a more intuitive and powerful way to model and compute with complex ideas.

MATHEMATICA AND WOLFRAM ALPHA: KEY APPLICATIONS

Two prominent applications stemming from Wolfram Language are Mathematica and Wolfram Alpha. Mathematica, launched in 1988, is a system built around Wolfram Language, primarily used for technical computations. Wolfram Alpha, on the other hand, is a computational knowledge engine that answers questions posed in natural language by converting them into computations within the Wolfram Language and querying its extensive knowledge base. These tools exemplify the language's ability to bridge raw data with structured computational understanding.

A BROAD AND DEEP FUNCTIONAL LANDSCAPE

The Wolfram Language is characterized by its immense breadth, containing approximately 6,000 primitive functions. These functions cover an astonishing range of capabilities, from manipulating boolean expressions and interacting with cloud services to processing discrete wavelet data and creating user interfaces. Examples like 'ImageIdentify' (which can recognize objects in images) and geographical functions (like finding nearest volcanoes and plotting them) demonstrate the language's power in integrating diverse real-world data and functionalities directly into the programming environment.

BUILDING THE WOLFRAM KNOWLEDGE BASE

The creation of Wolfram Alpha's knowledge base was a monumental, multi-decade effort. The methodology involved a brave approach to ingest vast amounts of information from the real world, treating it as a finite but extensive resource. Rather than starting with a grand theory, the process involved incrementally implementing knowledge across numerous areas, often with critical input from world experts in specific fields. This ensures a high level of accuracy and depth, aiming for expert-level knowledge across a wide spectrum of topics.

THE INTERPLAY WITH MACHINE LEARNING AND AI

While Wolfram Language focuses on knowledge-based computation, it increasingly integrates with machine learning. The goal is not to replace existing knowledge but to enhance it. Machine learning can now assist in exploring this vast computational knowledge base, particularly in areas that were historically difficult for computers, like image identification. The synergy between symbolic representation of knowledge and statistical machine learning methods is seen as a key path forward for more sophisticated AI capabilities.

FUTURE HORIZONS: SYMBOLIC DISCOURSE AND COMPUTATIONAL LAW

The long-term vision extends to a 'symbolic discourse language' that can computationally represent not just factual knowledge but also intentions, ethics, and complex human interactions. This includes applying computational principles to legal contracts ('computational contracts'), making them executable and unambiguous. The idea is to move beyond natural language ambiguity to a precise, computable form that computers can directly process, potentially revolutionizing fields like law, policy-making, and even personal decision-making through customizable ethical frameworks.

TRANSLATING CONCEPTS INTO COMPUTABLE FORMS

The endeavor to create a symbolic discourse language involves translating the actionable parts of conversations and common human concepts into precise, computable forms. While Wolfram Language can represent detailed nutritional data or the characteristics of fruits, it's still developing the representation for inherent human desires like 'wanting to eat.' This aspect is crucial for a complete system that can understand and act upon nuanced human intentions, extending beyond mere factual knowledge to encompass desires and motivations.

THE TURING TEST AND WOLFRAM ALPHA'S ROLE

Wolfram Alpha's capabilities offer a unique perspective on the Turing Test. While the test typically focuses on a machine's ability to exhibit human-like conversational behavior, Wolfram Alpha excels by providing factual, often obscure, knowledge. Its success in this area suggests it's not just mimicking human conversation but achieving a different, knowledge-based intelligence. This aligns with Alan Turing's initial vision of computational encyclopedias, indicating that Wolfram Alpha is a significant step towards a computational understanding of knowledge, potentially fulfilling the intent, if not the strict letter, of the Turing Test.

Mentioned in This Episode

●Software & Apps

●Companies

●Books

●Concepts

●People Referenced

Common Questions

Wolfram Language is a high-level computational language that focuses on symbolic representations of real-world concepts, unlike traditional languages that are more tied to intrinsic computer operations. It aims to be an abstract language from the beginning, allowing for direct manipulation of concepts like 'Boston' or spacecraft trajectories.

Topics

AI & Machine Learning Technology & Innovation Programming & Software Natural Language Processing Symbolic AI Computational Linguistics Future Of Computing Knowledge Representation Computational Contracts

Mentioned in this video

Software & Apps

Mathematica

A system that is an instance of Wolfram Language, first released in 1988, used for technical computations by typing computational language.

BooleanConvert

A function within Wolfram Language for converting between different types of boolean expressions.

GeoListPlot

A Wolfram Language function used to plot geographical data, demonstrated with the locations of nearest volcanoes.

Python

Mentioned as a comparison point for traditional programming languages, distinguishing Wolfram Language's symbolic approach.

ImageIdentify

A function in Wolfram Language that classifies the most likely object in an image, demonstrated with a webcam capture.

Wolfram Language

A computational language designed to work with symbolic representations of things that exist in the world or can be imagined, aiming to be the highest-level computer language.

Wolfram Alpha

A system for answering questions asked in natural language, which converts queries into computations using Wolfram Language and an underlying knowledge base.

SMP

The predecessor to Wolfram Language, designed by Stephen Wolfram as his first computer language, based on transformation rules for symbolic expressions.

GeoNearest

A function in Wolfram Language used to find the nearest geographical point of interest, demonstrated by finding the 10 nearest volcanoes.

Concepts

AI Ethics

A necessary module for automating content selection online, involving decisions about what content to block, up-rank, or show to users.

expert systems

Mentioned as an early AI approach from the 80s and 90s that focused on particular areas and is seen as having been too modest compared to the ambition of Wolfram Language.

Natural Language Understanding

A key problem in AI that requires broad world knowledge, which Wolfram Language and Wolfram Alpha aim to address by covering the whole problem rather than just pieces.

Principle of Computational Equivalence

A principle developed by Stephen Wolfram suggesting there is no bright line between intelligence and the computational capabilities of complex systems.

Turing Test

A test for artificial intelligence, discussed as a goal that Wolfram Alpha and Wolfram Language are arguably trying to solve, though it's not solely about mimicking human conversation.

People

Marvin Minsky

A well-known AI figure from MIT, who was shown a demonstration of Wolfram Alpha's question-answering capabilities and was impressed.

Gottfried Leibniz

A historical figure from the 1600s who had the idea of representing all of the world in a computational symbolic form, similar to Wolfram's concept of symbolic discourse language.

Elon Musk

Mentioned as someone who shares Stephen Wolfram's optimism and willingness to tackle daunting, seemingly impossible projects.

Locations

Boston

Used as an example of a symbolic thing that Wolfram Language can represent and work with.

Soviet Union

Mentioned in the context of historical struggles with ideologies and the potential for different AI systems to encode various ethical and ideological brands.

Companies

Google

Mentioned in the context of its 'dream' to make all information searchable and accessible, which Wolfram Alpha aims to build upon by adding understanding.

Twitter

Mentioned as a platform involved in automated content selection and the challenge of developing AI ethics modules for content ranking.

Facebook

Mentioned as a platform involved in automated content selection and the challenge of developing AI ethics modules for content ranking.

Books

Encyclopedia Britannica

Mentioned as an example of a comprehensive knowledge source that Turing considered making computational, a task Wolfram's team achieved more efficiently.

Organizations

Wikipedia

An example of a large aggregation of text that has not yet been converted into computable knowledge, contrasted with Wolfram's goal of computable knowledge representation.

finance

Bitcoin

Mentioned in the context of digital currency and ledgers, motivating the development of computational contracts.