The Biggest Shift Since The Industrial Revolution Is Here | Muriel Medard

Overview

Muriel Medard argues that AI is driving a permanent, world-scale shift in how information and work get done. Rather than competing with AI, she suggests blockchain and “ledger/proof” technology can serve as an important complement—especially as systems become increasingly automated and agent-driven.

Core themes and arguments

AI needs verification; blockchain can provide it

• Medard emphasizes that a major concern with today’s AI is the missing verification step: outputs can appear plausible, but it’s often unclear what data was used or whether the correct rules were followed.
• She argues that blockchain-based proofs/logs are incontrovertible by design, making them valuable as a “known-correct” layer that AI’s flexibility can build on.
• Together, AI (adaptation, action, speed) and blockchain/ledgers (auditability, correctness guarantees) can be “symbiotic.”

Blockchain can support AI by improving trust in data and agent actions

Medard highlights several concrete use cases:

• Provenance and verification of training data Ensuring models learn from legitimate sources rather than “junk.”

• Verifiable rule-following in agentic systems For example, confirming whether autonomous agents complied with constraints when executing transactions or other actions.

• Decentralized access to data streams So AI systems don’t depend on a single centralized source.

Key bottleneck: She notes that blockchain must become faster and more scalable, since waiting seconds for on-chain operations is too slow for many AI workflows.

Permanent economic/job shift, but decentralized systems may help people adapt

• In response to a claim (from a Wharton paper) that AI causes a permanent structural job shift, Medard agrees the change is permanent.
• However, she suggests blockchain could reduce some harms by enabling decentralized, democratized access to services, helping individuals profit from data and contributions in ways that are harder under centralized systems.

Quantum computing: crypto must focus on coding and “post-quantum” security

• Medard frames quantum progress as tightly linked to error-correction coding and information theory.
• For what crypto should do, she argues for shifting away from relying only on assumptions about computational hardness, and toward information-theoretic/coding-based approaches that remain secure even as quantum capabilities advance.
• She references post-quantum (PQ) security as part of this coding-centered response.

Optimum and agentic AI / data verification

• Medard describes Optimum as a “data acceleration” and “universal data acceleration” effort aimed at solving blockchain speed/bandwidth problems.
• She links Optimum directly to agentic AI needs:
  • Faster, verifiable acquisition of decentralized data streams (including data needed to package and propagate blockchain transactions).
  • Support for methods like federated learning and for fetching additional verification data so AI can confirm what it’s using.
• Her thesis: as AI agents transact more autonomously, demand grows for provable traceability and correctness, especially when there is no human “in the loop” to explain decisions after the fact.

Future of information gathering and human skills

• Over the next “50 years,” she predicts a progression in how information works:
  • from information as a push (broadcast media)
  • to information as a pull
  • to information that drives action based on a person’s intentions.
• On whether AI makes people smarter or dumber:
  • She cites a study-style argument where ChatGPT can improve initial creative output, but performance can regress when access is removed—while a control group that develops practice may improve more steadily.
  • Her takeaway: critical thinking and algorithmic/rigorous thinking become more important in an AI age.
• For young people preparing for durable careers, she encourages building judgment:
  • Learn math/algorithms and critical thinking to evaluate correctness and reasoning.
  • She argues against focusing only on coding as a language that changes over time.

Presenter / contributors

• David — host / interviewer
• Muriel Medard — Professor, MIT; co-founder of Optimum