The week's essentials in AI and exponential tech · 14 Sept – 21 Sept
The most consequential paper published this week on AI's systemic impact makes something explicit that most policy conversations still treat as secondary: the real bottlenecks in this transformation are not algorithms but semiconductors, data centre capacity, and energy infrastructure, and whoever controls those chokepoints controls the next redistribution of global power — this is precisely the argument I develop in Chapter 3, where I describe how the three forces of change compound each other in ways our linear intuitions consistently fail to anticipate. Our mind reads the curve as a straight line, and so we keep asking the wrong question — not "how fast is this moving?" but "is this really that different from previous technology waves?" — and by the time the answer is obvious, the lily pad has already covered half the pond. Layered on top of that structural shift, a new study shows that AI is not distributing its labour-market disruption evenly but amplifying pre-existing gender gaps, which connects directly to what I discuss in Chapter 12 as the adaptation crisis: the people with the fewest resources to adjust are absorbing the sharpest shocks. Meanwhile, research on programming students using ChatGPT finds that immediate performance improves while knowledge retention declines, a quiet trade that looks like a gain on every short-term dashboard and like a loss only when you zoom out — and zooming out is exactly what the exponential moment demands of us. And underneath all of this runs a detail that should unsettle anyone who believes governance can keep pace: the companies building world models, the most powerful predictive architectures yet attempted, are deliberately keeping their advances opaque, which, as I argue in Chapter 13, widens the gap between what the technology can do and what society can even see, let alone respond to. The problem, as always, is not technical. It is cognitive.
The paper reinforces the book's core thesis: exponential AI is not an isolated technical phenomenon but a catalyst for redistribution of economic, energy, and geopolitical power. Bottlenecks in semiconductors, data centre infrastructure, and energy create new forms of structural power that challenge the existing order.
Confirms a core thesis of the book: automation is not neutral. While AI exposure concentrates in higher-paid occupations (male-dominated), women in lower-wage sectors face displacement without equitable access to upskilling. A concrete example of how the adaptation crisis affects different groups unequally.
This finding exposes a core tension in exponential times: generative AI accelerates task completion but may weaken knowledge ownership and autonomous adaptive capacity. It challenges the assumption that higher short-term performance ensures future adaptability.
Competitive secrecy around AI training data and methods deepens the technological gap: only those controlling these secrets can operate at exponential scale, while others remain at the surface.
The question of whether the industry can or will slow its exponential pace exposes the gap between responsibility rhetoric and competitive reality. This embodies the core challenge of governing disruption without halting innovation.
If this week helped, pass it on to someone else living these times.
Create your free account and get Exponential Times every Monday: the news interpreted, threaded through the book that explains it.
Privacy
We use necessary cookies for the interactive book to work, and we'd also like to use anonymous analytics to improve it. Your reflections are always private and we never analyse them. More details in the cookie policy.