One paradox. Five clues. First observed in the 19th century. Now reshaping the AI revolution.

Logic says that making something more efficient should reduce how much of it we use. But economists discovered the opposite can happen. As AI models become smaller, faster and cheaper, the world’s total demand for computing is exploding instead of shrinking. Can you name this paradox before the final clue?

Clue #1 — Making something cheaper can make people use far more of it

Most people assume efficiency automatically saves resources.

Economics says otherwise.

When a technology becomes more efficient, its cost usually falls. Lower prices encourage wider adoption, new applications and heavier usage. In many cases, total consumption rises instead of falls.

What begins as an efficiency gain can end as a surge in demand.

Clue #2 — It was first discovered while studying Britain’s coal boom

This idea dates back to 1865, when a British economist noticed something unexpected.

Steam engines became far more efficient at burning coal. Instead of reducing coal consumption, Britain used even more coal, because cheaper energy powered more factories, railways and industries.

Greater efficiency made the resource more valuable—not less.

A prediction from the Industrial Revolution is now reappearing in the AI Revolution.

Clue #3 — Today’s AI boom is becoming the world’s biggest real-world example

Throughout 2025–2026, companies released increasingly efficient AI models using techniques such as quantisation, pruning, sparse architectures and Mixture-of-Experts (MoE).

Each query now requires less computation than before.

Yet AI demand is growing even faster.

Search engines, office software, smartphones, customer support, education, coding, healthcare and scientific research are embedding AI into billions of daily interactions. Global investment in AI infrastructure and hyperscale data centres has accelerated as companies race to meet exploding compute demand.

Cheaper intelligence is creating far more intelligence.

Clue #4 — The world’s energy demand keeps rising despite smarter chips

Modern AI chips deliver far better performance per watt than previous generations.

However, technology companies continue announcing multi-gigawatt AI campuses, while electricity demand from data centres is projected to rise sharply over the coming decade.

The reason is simple:

Every efficiency breakthrough reduces cost…

…which encourages millions of new users, products and AI-powered services.

Saving energy per task does not necessarily reduce total energy use.

Clue #5 — Its name comes from the economist who first described this counter-intuitive effect

This principle carries the surname of a 19th-century British economist.

Today it appears in discussions on AI, electric vehicles, cloud computing, renewable energy and climate policy.

“Many economists also refer to it as the “rebound effect” (when the increase is under 100%) or “backfire effect” (when it exceeds 100%).”

Its message is straightforward:

Efficiency alone cannot guarantee sustainability unless demand also remains under control.

So — what is this paradox?

It explains why technologies that become more efficient can sometimes consume more resources overall because lower costs trigger greater adoption.

First described during Britain’s coal age.

Now central to debates on artificial intelligence, energy security, climate policy and the future of global computing.

As AI becomes cheaper, faster and more efficient, this paradox is becoming one of the defining economic ideas of the 21st century.

Bonus — can you name:

• The British economist who first described this paradox in 1865
• The alternative name commonly used for this phenomenon
• The resource whose increasing efficiency inspired the original theory
• One AI technique (such as Mixture-of-Experts or quantisation) that improves efficiency but can contribute to this paradox through wider adoption

Drop your answer below. Unlike Wordle, solving this one could help explain why smarter technology doesn’t always mean lower resource consumption. Day #66 arrives tomorrow.

Missed Yesterday Challenge?