Genius Makers
This is no longer on the current shelf — shelves rotate as new material clears the bar. The analysis below is unchanged. See what is featured now.
A gripping narrative history of how the AI race became the defining competition of our era, told through the ambitions and egos of the people who built it.
What it covers
Genius Makers traces the modern AI revolution from Geoffrey Hinton's breakthrough in deep learning through the fierce competition between tech giants to dominate the field. Metz weaves together personal stories of researchers, entrepreneurs, and executives—from Hinton and Yann LeCun to the corporate strategists betting billions on AI—showing how academic breakthroughs became industrial superpowers. The book covers the evolution from neural networks to transformers, the talent wars, and the geopolitical stakes that emerged as AI became central to tech dominance.
Why it matters
In 2026, understanding how we got here is crucial—the decisions made in the labs and boardrooms of 2015-2020 directly shaped the AI landscape you're building in today. The book illuminates why certain architectural choices won, how talent concentration shaped the field, and the often-overlooked role of infrastructure and compute in determining winners. For anyone building agentic systems or evaluating models, knowing this history reveals the path dependencies and power dynamics still shaping what's possible.
What it made me think
Reading this reminded me that AI progress isn't inevitable—it's the result of specific bets, personalities, and institutional choices. It's like watching a chess game where the opening moves (Hinton's 2012 breakthrough, the GPU revolution) cascaded into the entire game we're playing now. It made me realize that the 'obvious' winners of today weren't obvious in 2015, which should humble us about predicting what's next.
The short version
The AI revolution wasn't written by algorithms—it was written by people making specific bets at specific moments, and understanding those bets explains why your options today look the way they do.
My take
What struck me most is how much of the 'genius' in the book was actually about recognizing when a bet was going to pay off and having the conviction to go all-in. When I think about building agentic systems or evaluating production AI, that same principle applies—you're not just picking the best model, you're predicting which architectural choices and inference strategies will compound. The people who won in the 2010s weren't smarter; they were better at reading the direction of travel and positioning accordingly.
How it connects
- The talent wars Metz describes directly explain why evaluation and benchmarking became so critical—when you can't hire all the genius, you need better ways to measure what's actually working
- The compute concentration he documents is why model compression and efficient inference (MoE, quantization) are now existential for anyone not backed by infinite resources
- The geopolitical dimension explains the current push toward open models and research transparency—it's a counterbalance to the centralization he chronicles
Bottom line
Read this not for nostalgia, but to recognize the power structures still shaping AI in 2026 and see where the cracks are forming.
Takeaways
- Talent concentration matters enormously—the researchers who moved between academia and industry shaped entire research directions and competitive outcomes
- Infrastructure (GPUs, datasets, compute) was often the hidden advantage that determined who could push boundaries, not just raw algorithmic insight
- The transition from academic publication culture to corporate speed created a permanent shift in how AI research gets done and who controls it
- Geopolitical competition became real earlier than most people realized—by 2017-2018, AI was already a strategic asset, not just a research frontier
- The book reveals how much of 'genius' is actually persistence, access to resources, and being in the right place when a breakthrough becomes possible
Brendon Score: 7.3/10
- Relevance: 8.0/10 — +2.00
- Depth: 8.0/10 — +2.00
- Actionability: 6.0/10 — +1.50
- Freshness: 7.0/10 — +1.75
- Average: 7.25
- Total (rounded): 7.3/10