A large collection of cars, buses, cabs, and pedestrians are stuck in a traffic jam that stretches in multiple directions at an intersection in New York City.

Why does it take so long for science to spot harmful technology products and support the search for safer alternatives? And how can we speed it up?

This week when I opened my box of summer clothes after a long Ithaca winter, I wondered what to do with a jacket I’ve been avoiding for several years. The problem? It’s coated with toxic PFAS that are linked with cancer and other health issues. Chemical companies marketed and distributed these substances for decades to the point where they are now in the bloodstream of 98% of Americans. Despite internal corporate studies from the 1960s that the chemicals could be poisonous, these risks were not more widely known to science until the late 1990s. Thanks to lawsuits, regulation, and an overhaul in the chemicals industry called “Green Chemistry,” companies are now starting to produce safer alternatives over 60 years later.

I started reading the history of PFAS last year while wrestling with the slow pace of science on digital technology harms, a conversation I’ve been having with Dr. Amy Orben, a psychologist at the University of Cambridge and the UK Medical Research Council’s Cognition and Brain Sciences Unit. Amy is one of several researchers caught up in what Holden Thorp calls the “unsettled science” of social media and mental health: a high-stakes debate involving families, corporations, and scientists trying to reconcile the evidence. Last year, Amy reached out to discuss an even more fundamental question: how did we get into this mess, and how can science do better in the future on social media, AI, and other emerging digital technologies?

Amy and I have now published an article in Science on “Fixing the science of digital technology harms” that proposes an answer, building on lessons from the history and philosophy of science, statistics, and the commercial determinants of health. For high-stakes issues where companies control the capacity to do research, firms are strongly incentivized to use science as a tool for delay. Just like the 20th century chemicals industry, these incentives motivate tech leaders to launch products with half-working safety systems and require the highest standard of proof before investing in safer alternatives. Our article offers four ways to break out of this pernicious cycle.

The traffic jam in science

Here’s the dilemma: it would be disastrous to lower scientific standards, but society also often can’t wait for the highest standards of scientific evidence to be located—especially since the necessary data and funding are not provided in a trustworthy manner. The result is a traffic jam that fails to serve the public interest. This pile-up can accumulate harm and bottle up scandals until they burst in ways that helps no one.

Here’s how the traffic jam happens. High standards of evidence slow down the search for public safety when we use science as a gate-keeper: setting a high threshold of evidence before searching for safer alternatives. In the over-simplified version of this (illustrated below), scientists only start studying something after significant public concern. Then we only start testing interventions until after the harm is proven and shown to be widespread. Next, scientists advise that interventions for safer technology be adopted only after further testing, efforts that are hindered by the tech industry’s stranglehold on research. 

With each step, scientists are making principled decisions under the circumstances. But from the bird’s eye view, that’s how 60 years can pass before PFAS are replaced and how decades can pass between the first voiced concerns about social media and our current intense debates on what to do about it.

Our article in Science is only four pages, so I encourage you to read it in full. In the article, we make four suggestions for solving this traffic jam. The first two suggestions are very simple, practical things that can be done right away:

  1. Take tech companies out of the driver’s seat of science: accelerating research with better incident reporting of harms from the public and better funding for independent research, potentially sourced from taxes on the tech industry.
    • This is where CAT Lab’s focus on participatory science has an important role to play. As we write in the article, “We gain nothing when people are told to mistrust their lived experience owing to an absence of evidence when that evidence is not being compiled.”
  2. Develop better ethics and scientific standards for communicating the state of science and weighing the risk of being wrong: after all, business losses are more easily reversed than deaths

These ideas can only go so far. They try to solve a traffic jam by driving smarter rather than re-designing the roads. In the second half of the paper, we propose two ways to fix the deeper problem: a fast lane and a priority list for safety science.

A fast lane for safety science

If the science of technology safety is like a one-lane road, an obvious solution is to parallelize the search for safer technologies. Drawing from work by the philosopher Jacob Stegenga on “Normal” and “Fast” science, we suggest a “Minimum viable evidence” model of evidence generation. In this model, we still rely on science to confirm that individual harms are happening (stage 1), study how prevalent those harms are (stage 2), and test policy interventions (stage 3) before scaling them. But wherever a concern is especially high-stakes and companies are especially unsupportive of science, scientists and policymakers should adjust our threshold of evidence for each of these stages to work more quickly in parallel.

Importantly, corporate obstruction should motivate us to accelerate scientific concern and the search for evidence on safety. As we write: “if technologists deploy minimum viable products without adequate testing or support for independent research, their products will be subjected to adverse inferences concerning safety.”

Prioritizing the search for safer technologies

One reason tech companies obstruct the science of safety is that it’s not clear to them how science will increase their return on investment. Investors and executives are understandably incentivized to resist a system that is focused on detecting harmful products and penalizing them for the consequences. And when companies dig in their heels, it can take decades for safer alternatives to be found. That’s what scientists, policymakers, and the chemicals industry have been trying to change through the idea of “Green Chemistry,” where innovators are incentivized to prioritize safer products.

Here’s how it works in the state of California: a central body maintains a list of substance uses that are potentially toxic, based on data submitted from many different participating organizations. Presence on the list doesn’t mean that something’s harmful – it just means that there’s a possibility of harm. Scientists can review the list when deciding what to test next, and innovators can use the list to identify the economic potential of safer alternatives. Then, if scientists do find that a substance is harmful, there might already be a viable alternative. When industry adopts that safer alternative, the innovator who prioritized safety will get a return on their investment.

If designed well, this market-based approach would support a continuous search for safer alternatives as the technology industry evolves. The maintainers of the priority list could continue to add to it as new technologies become widespread, including AI and other emerging systems.

Coordinating Effectively for Safer Digital Technologies

Both the fast lane for safety science and the priority list only work with coordination. As we conclude in the article:

History will already judge science for its ability (or inability) to recognize an evidence creation and utilization system that cannot deal with challenges that will only magnify further in the age of AI. It is up to scientists and policy-makers to face up to the failings of this system, and to help craft a new one that will allow science to be the engine—not the brake—to creating a positive digital society for decades to come.

If you’re a scientist, regulator, consumer protection advocate, or industry leader working on related issues, thank you for the work you are already doing. Many people and organizations already hold parts of the solution, and we’re eager to support efforts that are already underway. Dr. Orben and I will be making several announcements with others in the coming months about next steps, and we are looking for collaborators. So please get in touch!

Image source: New York City Gridlock, on Wikimedia Commons.