This past Spring the Environmental Protection Agency announced the first ever national drinking water standard for per-and polyfluoroalkyl substances (PFAS). PFAS are more colloquially known as “forever chemicals,” a moniker they’ve achieved because of their strikingly persistent ability to resist degradation whether that be from water, air, heat or anything else that might break down biodegradable compounds. These compounds live forever (or at least 1000 years according to some studies) in landfills, in our water, and in our bodies. While PFAS levels vary widely from person to person, it's generally accepted that they are detectable in the blood of 95% or more of the population in industrialized countries.
Regulations to eliminate toxic chemicals like PFAS are needed. They are also far too limited in scope, far too expensive to enact in relation to their reach, and by placing responsibility on local governments, they are setting up a plan of perpetual clean up, instead of eliminating the chemicals at the source.
Consider this. The new regulations targeting drinking water, while a legitimate step forward, cover only two of the 15,000 known PFAS chemicals for which the US maintains a database. The two covered in the ruling are the most well known, and most studied PFAS chemicals – PFOA and PFOS. Why only two? Because studying 15,000 chemicals in isolation is an overwhelming task for the government. These particular chemicals occur in complex formulations in tens of thousands of everyday and industrial products. For companies using them, studying them in product formulations would be a herculean task, involving first identifying them in thousands of formulations, and then evaluating, prioritizing the products that should be reformulated first, potential replacement ingredients, and how those would behave in combination with other chemicals in their specific product formulations.
Right now the onus is on local governments to get these two PFAS out of drinking water, the cost of which is estimated at $1.5 billion annually, a bill for which responsible parties are not clamoring to pay at the moment. I would argue that it’s in the best interest of consumers, governments and even corporations to tackle these chemicals at the point of making decisions about product formulations. This no longer has to be a complex, costly and time consuming task because AI is making that process possible now.
AI for Science is progressing incredibly rapidly and already enabling scientists to run thousands of experiments in software. Such systems that analyze complex datasets to identify toxicity of compounds and cohesion with existing formulations can replace individual costly and time consuming lab based experiments. It accomplishes this by training an AI model with relevant scientific knowledge, understanding of physical systems, and relevant data sets and restricted substance lists. Once this is done, scientists can then quickly explore the vast space of possible chemical formulations or combinations, identifying compounds that could replace “forever chemicals” and other toxic ingredients without sacrificing performance or quality.
AI models can also help product developers evaluate the trade-offs between different ingredient alternatives in terms of cost, availability, safety, or environmental impact, and now approaches are emerging that provide valid predictions even for the very small datasets typically available to chemicals and materials companies. This predictive ability makes it possible for companies to identify replacement chemicals in formulations more easily and more cost effectively, enabling compliance with regulations a proactive, rather than defensive move. This will also result in scientists having more time to work on the most critical experiments.
In an era when AI models are making it possible for all companies involved in the chemicals and materials space, as well as regulatory agencies and groups, to more quickly research and evaluate chemical compounds, we no longer need to move at a snail's pace. The need for further regulations – will not only enhance consumer trust in manufacturers, but lead to a healthier planet, an overall win-win situation.
Sunil Sanghavi is CEO of NobleAI, a pioneer in science-based AI solutions. Previously, Sunil was a Senior Investment Director at Intel Capital, where he invested in AI/ML hardware and software companies. He attended the Indian Institute of Technology Bombay and obtained a BSEE from Cal Poly, San Luis Obispo.
