Stephen Wolfram: Physicist, computer scientist, and founder of Wolfram Research. Creator of Mathematica, Wolfram Language, and Wolfram Alpha, and author of 'A New Kind of Science.'
What this episode covers
Stephen Wolfram describes Wolfram Language as the highest-level computer language ever built, a symbolic language whose primitives represent things that exist in the world rather than raw computer operations. He walks through live demos like image identification and finding nearest volcanoes, explaining the decades-long effort to build the Wolfram knowledge base behind Wolfram Alpha. The conversation moves into AI history, the relationship between machine learning and curated computational knowledge, and the vision of a 'symbolic discourse language.' Wolfram discusses computational contracts, automated content selection and AI ethics, and proposes a market of differing ethical modules people can choose. He closes reflecting on optimism, long-tail projects, the Turing test, and Alan Turing's own dream of computational knowledge.
The guest's own work
Wolfram Language
Stephen Wolfram
“what is Wolfram language in terms of sort of I mean I can answer the question for you but is it basically”— Stephen Wolfram
Mathematica
Stephen Wolfram
“came out 1988 it's this system that is basically a instance of Wolfram language and it's used to do computations”— Stephen Wolfram
SMP
Stephen Wolfram
“I designed the predecessor of what's now often language the thing called SMP which was my first computer language”— Stephen Wolfram
Wolfram Alpha
Stephen Wolfram
“this kind of with the Wolfram language along with the Wolfram Alpha represents kind of what the dream of what AI is supposed to be”— Stephen Wolfram
Wolfram Engine
Stephen Wolfram
“we have you know this free Wolfram engine for developers which is a free version for developers”— Stephen Wolfram
A New Kind of Science
Stephen Wolfram
“I worked on my new kind of science project instead of exploring the computational universe and came up with things like this principle of computational equivalence”— Stephen Wolfram
Also referenced (named, not recommended)
BERT
“I think you have some of the state-of-the-art transforms okay have Bert in there I think oh you know so”— Lex Fridman
Big reveals from this episode
- Wolfram claims Wolfram Language is 'by far the highest level computer language that exists,' built deliberately opposite to other languages.
- Live demo: image identify misclassifies Wolfram as a plunger before getting 89% 'person' on a better photo.
- He admits surprise Wolfram Language isn't more popular and blames slow idea-absorption plus his focus on products over commercialization.
- Argues open-sourcing everything would NOT work because maintaining knowledge integrity requires centralized leadership.
- Reveals he initially assumed building Wolfram Alpha would require solving general AI, then bet his own computational-equivalence paradigm could shortcut it.
- Story of Marvin Minsky testing the system with random medical, chemistry, astronomy queries and concluding 'oh my god it actually works.'
- Argues there must NOT be a single AI ethics module governing the world; proposes multiple value-system-branded modules people self-select.
- Says Wolfram Alpha bots lose the Turing test instantly because they answer obscure knowledge no human would know.
Worth remembering
- Wolfram Language has around 6,000 built-in primitive functions covering hugely varied domains.
- The language natively knows every volcano in the world and can compute and map the nearest ones.
- Wolfram Language is 33 years old, existing for more than half the entire ~60-year history of computer languages.
- Mathematica/Wolfram Language is effectively free via site licenses at essentially all major US universities.
- Wolfram took his early team to a big reference library and set the goal of ingesting everything in it within a year or two.
- His principle: any knowledge area lacking a real expert helper wouldn't come out right.
- Notes Gottfried Leibniz proposed computational contracts in the 1600s but was ~300 years too early.
- Alan Turing considered making Encyclopedia Britannica computational and estimated the effort, being more pessimistic than reality.
- Computational contracts could encode rules like what a self-driving car should do, written in computable rather than legalese form.