Inside a plan to use AI to amplify doubts about the dangers of pollutants

An industry-backed researcher who has forged a career sowing doubt about the dangers of pollutants is attempting to use artificial intelligence (AI) to amplify his perspective.

Louis Anthony “Tony” Cox Jr, a Denver-based risk analyst and former Trump adviser who once reportedly claimed there is no proof that cleaning air saves lives, is developing an AI application to scan academic research for what he sees as the false conflation of correlation with causation.

Cox has described the project as an attempt to weed “propaganda” out of epidemiological research and perform “critical thinking at scale” in emails to industry researchers, which were obtained via Freedom of Information Act requests by the Energy and Policy Institute, a non-profit advocacy group, and exclusively reviewed by the Guardian.

He has long leveled accusations of flimsiness at research linking exposure to chemical compounds with health dangers, including on behalf of polluting interests such as cigarette manufacturer Philip Morris and the American Petroleum Institute – a fossil fuel lobbying group he has even allowed to “copy edit” his findings. (Cox says the edit “amounted to suggesting a small change” and noted that he has also obtained public research funding.)

Both the tobacco and oil industries have a history of weaponizing scientific uncertainty, experts say, with some arguing that similar tactics drive the Trump administration’s current deregulatory efforts. The president’s May “gold standard” science order, for instance, empowered his appointees to “correct scientific information” and “discipline” those who breach the administration’s views, prompting outrage from some scientists.

Cox has obtained funding to develop the new AI reviewer from the American Chemistry Council (ACC), the nation’s largest chemical industry advocacy group, which counts oil and chemical giants such as Exxon and DuPont as members.

Experts say the ACC’s sponsorship raises questions about whom the project will benefit.

Asked about these concerns, Kelly Montes de Oca, spokesperson for the ACC, said: “This research has the potential to support scientific understanding and analysis of chemical exposure and human health, enhance transparency and reproducibility, advance the safety of chemical products and processes, and inform science-based global regulatory approaches.”

Cox said in an email to the Guardian that his assistant “is specifically designed to be helpful to those who wish to understand the objective implications of data without any distortions from the kinds of well-known human heuristics and biases that make objective analysis difficult for humans”.

“My work aims to help anyone interested in using sound technical methods to pursue scientific truth,” he added. The questions sent to him by the Guardian contained “many fundamental inaccuracies”, he said.

Some critics have mischaracterized my work as an attempt to delay regulation or promote industry interests. That is not true - Louis Anthony ‘Tony’ Cox Jr

Cox said the tool is currently being tested on submissions to academic journals – including Risk Analysis, which he edits – to evaluate research submissions before they are submitted to the peer review process.

Asked for a response to concerns about the project’s funding, Cox said that he has publicly acknowledged the ACC’s support in all relevant publications and said the tool “has no axe to grind and no positions to push”.

But the ACC is not a neutral force, said Chris Frey, the associate dean for research and infrastructure at the North Carolina State University’s College of Engineering who chaired the Environmental Protection Agency’s clean air scientific advisory committee from 2012 to 2015.

“They lack impartiality in that they want to minimize regulatory burden on their members,” said Frey.

ChatGPT

In mid-February of 2023, Cox struck up a conversation with the AI assistant ChatGPT. He later sent the chat to his University of Colorado email address,which is subject to public record request laws. 

After asking the chatbot to write a sonnet about Abraham Lincoln, he turned the conversation to a more serious topic: the tiny toxic particles known as PM2.5, also known as fine particulate matter.

Scientists have long found links between PM2.5 exposure and a wide variety of health concerns, from asthma and lung disease to heart attacks. But for years, Cox has raised uncertainty about those connections, publishing work “re-assessing” the mortality risks associated with exposure to the pollutantin animal agriculture and challenging the link between PM2.5 from gas stoves and childhood asthma.

On that February day, Cox asked ChatGPT a simple question: “Does PM2.5 cause lung cancer?” The bot responded by noting the “strong scientific evidence” demonstrating the link.

A seemingly dissatisfied Cox went on to ask if it is “known with certainty that PM2.5 causes lung cancer”, calling on ChatGPT to identify potential confounding factors in the research on the air pollutant. He took brief tangents to pose riddles to the bot and to request additional sonnets – including one about PM2.5 itself, yielding the line “no nose can catch you, for you are so fine”.

Eventually, Cox asked ChatGPT whether the association between PM2.5 and lung cancer can be explained by a “combination of residual confounding, measurement errors, and modeling biases”. The bot maintained that “the observed association is likely to be real and not fully explained by these factors”.