Analyzing the News: The Nobel Prize for optogenetics

Twelve outlets covered this year's Nobel Prize in medicine. Fíor scored eleven of them and read four as AI.

Fíor is a natural language classifier that predicts whether a piece of text was more likely written by a person or by AI.

On 5 October, the Nobel Prize in Physiology or Medicine went to Karl Deisseroth, Peter Hegemann and Georg Nagel for optogenetics, a way of switching individual nerve cells on and off with light. Hegemann and Nagel found the key protein in a single-celled alga, and in 2003 showed it could work in other cells. Deisseroth showed it working in rat nerve cells in 2005. The three share 12 million Swedish kronor.

The technique is now a standard tool for studying how particular brain cells shape behaviour. Treatments based on it are still being researched, and several outlets were careful to say so.

The outlets were Reuters, AP, CNN, the Guardian, the Washington Post, Le Monde, Euronews, the BBC, The Indian Express, Hindustan Times, The Hindu and WBUR. That is two wire services, outlets from the US, Britain and France, three Indian national newspapers and a US public radio station. The BBC scan was of a copy hosted on a web address that is not the BBC's own, so it is left out. AP and CNN were scanned through republished copies. WBUR's piece is a transcribed interview with Deisseroth, scanned as republished by Alabama Public Radio. The Hindu's piece is a short summary that ran eleven days after the announcement.

The result

Seven read as human: Reuters, AP, CNN, WBUR, The Hindu, The Indian Express and Hindustan Times.

Four read as AI: the Guardian, the Washington Post, Le Monde and Euronews.

None was a close call, though Le Monde leaned AI less strongly than the other three.

This is the fifth story in the series, and some outlets keep landing the same way. Reuters has read as human every time, from the IMF forecast to the AI sell-off and now this. The Guardian has read as AI in all four pieces where it ran its own reporting. Euronews has read as AI all three times it has appeared.

Others move. AP and CNN both read as AI on the AI sell-off and as human here. Five stories is still a small base.

The ones that read as AI

All four were science explainers. Each walked the reader from a light-sensitive protein in pond algae to a switch for the brain.

Explaining science is the obvious pattern. It does not hold up on its own. The Indian Express and Hindustan Times also ran explainers on the same science, and both read as human.

The human side leans the other way. Reuters, AP and CNN built their stories around the announcement, with quotes from the committee and the laureates. WBUR's piece is mostly Deisseroth explaining his own work in his own words.

Why science coverage reads like this

The Nobel committee publishes its own plain-language account of the science on the day of the announcement. Explainers written against the clock start from it, and most followed the same sequence: the alga, the protein, the switch, the brain.

Some of these pieces may have had AI help, with a first draft, a rewrite or a final polish. Or science explainers have become so formulaic that a reporter and a model would write them the same way. Fíor cannot tell those apart on its own.

The prize rewards work that took the best part of twenty years to prove. Most of the explanations of it were written in a morning, from the same announcement.

What Fíor is

Fíor is a machine learning classifier, currently in beta and available as a browser extension. It gives a prediction, not a verdict. It cannot see who sat at the keyboard, and nothing here is a claim about any journalist or publication. Reading as AI and being written by AI are not the same thing.

Treat it as a signal, not proof, and check the pages you read yourself.

Try the browser extension at Fíor.

Analyzing the News is a series where we run one major story through Fíor and see what comes back.