AI is getting a medal for solving it

And our top AI story of the week

Author

AI For All
November 21, 2025

In partnership with

Hello readers,

Welcome to the AI For All newsletter! Today, we’re talking about AI’s charge into the world’s most challenging mathematics, IT infrastructure in the age of AI, and more!

AI in Action: Figuring it out with the best of them

AI has long excelled at games like chess and Go, but struggled with something as rigorously human as advanced mathematics. That may be changing. DeepMind’s new AI system, AlphaProof, recently matched the performance of silver medalists at the 2024 International Mathematical Olympiad. But the real breakthrough is less about the score and more about how the AI got there.

AlphaProof operates within a formal math environment called Lean, which checks whether proofs are logically valid. To overcome a lack of training data, DeepMind trained a large language model to convert natural-language math problems into Lean format, generating 80 million examples. From there, AlphaProof used reinforcement learning and a tree search algorithm to efficiently explore valid proof paths, just like AlphaZero did with Go.

Its most innovative feature, Test-Time Reinforcement Learning (TTRL), let AlphaProof generate easier variations of tough problems, solve those, and apply the insights back to the original challenge, a clever parallel to how humans learn. This allowed it to crack Olympiad-level problems that defeated most participants.

AlphaProof still has limitations: it needed help on geometry questions, consumed huge computational resources, and isn’t ready to invent new math. But it offers a glimpse of AI not just solving known problems, but helping humans push the boundaries of mathematical discovery.

Voice AI Goes Mainstream in 2025

Human-like voice agents are moving from pilot to production. In Deepgram’s 2025 State of Voice AI Report, created with Opus Research, we surveyed 400 senior leaders across North America - many from $100M+ enterprises - to map what’s real and what’s next.

The data is clear:

  • 97% already use voice technology; 84% plan to increase budgets this year.

  • 80% still rely on traditional voice agents.

  • Only 21% are very satisfied.

  • Customer service tops the list of near-term wins, from task automation to order taking.

See where you stand against your peers, learn what separates leaders from laggards, and get practical guidance for deploying human-like agents in 2025.

🔥 Rapid Fire

📖 What We’re Reading

How to Optimize IT Infrastructure in the Age of AI

Artificial Intelligence (AI) and Machine Learning (ML) technologies have completely revolutionized IT infrastructure. Whether they’re used to create predictive analytics for real-time resource management and anomaly detection or for task automation to optimize operational efficiency, these technologies are propelling the IT field into an era of unmatched innovation, speed, scalability, and efficiency. In this fast-paced, competitive environment, IT leaders must have in-depth knowledge of these technologies’ functions and applications as well as concrete strategies for implementing them into existing infrastructures.

Srinivasa Raju Pakalapati is a seasoned DevOps leader who specializes in AI-driven infrastructure and automation. In this Q&A, Pakalapati explains what it takes to keep pace with the rapidly advancing digital world and shares strategies for overcoming its challenges.