Stephen Wolfram answers questions from his viewers about the future of science and technology as part of an unscripted livestream series, also available on YouTube here: https://wolfr.am/youtube-sw-qa

Questions include: What, exactly, is an "AI agent"? "Agentic"? It seems like nobody knows what those words actually mean today. - Can you tell us about the future of media/information consumption? Will we become a society of "AI summaries" as our main form of information gathering? - Before AI summaries, there were encyclopedias and textbooks and CliffsNotes and such, and while they were useful and convenient, they never became de facto. - When will we get the first AI/robot news reporter? I see these being useful in cases of dangerous live broadcasting like hurricanes, to keep people up to date. - How far are we from LLMs generating a Stephen Wolfram–style long-form post, with similar elucidations, based on a short prompt of the key insight or topic? - When you say the teaching is delegated to the machine, are you saying that the machine is telling the student what to think about instead of just answering questions? - Can a sentient AI "understand" how humans learn? If we would delegate to them the teaching of human kids, would that be compatible with a biological point of view? - Have you ever considered entering the robotics space? A Wolfram Robotics, so to speak? - But if people delegate all calculations to the machines, then might it not happen that the machine actually learns to ask better questions than the humans can, since the machines have the experience built from the calculations and the humans don't? - What will AI not be able to do? Do you believe that something like that exists? - Tiny humans care about those questions about clouds and trees. - Robotic trade shows sound interesting. The company Boston Dynamics shows a lot of progress in the humanoid department. - Anything to say about the future of pi? (Happy Pi Day!) - ​​Do you expect LLM development to hit significant diminishing returns within the next 2–3 years? - Automated theorem proving is so interesting. I'm trying to figure out how to make a theorem prover that demonstrably collapses a/the wavefunction. Like Stephen said; quantum LLMs.