DEPARTMENTS
PRESIDENT'S MESSAGE
NICOLE GREESON, MS, CIH, is associate director of the occupational and environmental safety office at Duke University and Health System.
Send feedback to The Synergist.
AI’s Implications
BY NICOLE GREESON, AIHA PRESIDENT
This issue of The Synergist includes a report on AIHA’s most recent environmental scan, which identified 76 trends that affect our profession. You can read more about the scan in the report (PDF). Here, I’d like to focus on the 12 trends related to artificial intelligence and large language models (LLMs). These will be of interest to the many OEHS professionals who are trying to understand the implications of AI for their work:
Too much trust. The media has publicized examples of inappropriate AI use, such as a lawyer who used a chatbot to research a case and received a report with fake legal precedents. The challenge for the OEHS profession is to create guidelines for ethical, appropriate use of AI tools.
Replacing search. Internet searches powered by AI result in summaries that are presented more prominently than links to webpages and documents. This change means that AI tools will become important sources of health and safety information for workers and the public.
Smarter robots. The integration of AI with robotics may result in the development of robotic systems that can fulfill more roles in the workplace. But the unpredictability of AI models means that we will need to address the potential hazards presented by these systems.
Science without theory. AI tools are “black boxes”: nobody knows precisely how they work. As this technology proliferates, so too will the black-box problem. One possible result is that more solutions in science will be based on correlation, not causal theory.
Automation of science. AI systems will be able to perform some kinds of experiments more quickly and at larger scale. Their ability to write code may lower barriers to entry for some scientific fields.
Domain-specific AI models. Companies are creating proprietary AI tools trained on private data, but these smaller training datasets increase the likelihood of inaccurate output.
We will all feel the influence of AI tools for the foreseeable future.
IH data standardization. AIHA has long sought to create a common standard for the collection, storage, and sharing of occupational exposure data. In the era of big data and AI, the need for standardization is even greater.
Enterprise-scale LLMs. Technological developments that allow AI systems to be deployed on desktop computers may help alleviate the privacy concerns related to sharing data on cloud-based servers. We may be asked to integrate our OEHS data with our company’s internal LLM.
Tools for chemistry. Researchers are developing an AI tool that can discover or design the materials for pharmaceuticals. Such a tool has potential for harm, so researchers have built in safeguards, such as using the Globally Harmonized System of Classification and Labelling of Chemicals to check for explosive molecules.
AI and responsibility. As AI becomes more prevalent—for example, through increasing automation in factories—workers may be held responsible for any harm resulting from flaws in the design of AI tools.
AI devising dangerous materials. The potential exists for AI to provide instructions for making chemical weapons or mutations that improve the transmission of disease.
Impact on cybersecurity. AI is increasing the ability of hackers to wreak havoc.
Some of these trends may directly affect our work, and others will likely have only indirect impact. But clearly, we will all feel the influence of AI tools for the foreseeable future.
AIHA is already working to get ahead of these developments. In May, we published an AI responsible usage policy (PDF) intended to ensure that members and staff use AI ethically and in alignment with AIHA’s mission and values. A forthcoming white paper on IH data standardization was nearing completion as this issue of The Synergist went to press. Another volunteer effort led by AIHA Board member Bradley King is developing a guidance document that addresses the challenges, risks, and benefits of AI to OEHS professionals. We’re hopeful that this document will be available in the first half of 2025.
As we continue to sort through the implications of AI, we want to hear what you’re experiencing. Please share your input with us.
RESOURCE
Foresight Alliance: “AIHA Environmental Scan: Trends Affecting OEHS” (PDF, January 2024).