EHS Vendors Must Embrace The Open-Source AI Revolution To Keep Pace In The Product Development Race

  • Blog
  • Environment, Health & Safety

EHS Vendors Must Embrace The Open-Source AI Revolution To Keep Pace In The Product Development Race

While EHS technology buyers may not be wholly certain of what to expect from AI, a recent Verdantix survey showed respondents are anticipating it have a revolutionary effect on EHS. Several AI use cases for EHS have been developed into commercially viable solutions over the last four years. Much of the development has been driven by specialist vendors, for instance, Intenseye and ProtexAI have applied computer vision to collect video data of incidents, near misses and safety violations. Libryo (acquired by ERM) and ehsAI (acquired by Intelex) leverage natural language processing (NLP) to streamline compliance and permitting. Benchmark Gensuite has been a notable frontrunner in AI adoption in the EHS management software space: its Describe-It AI advisor tool assists frontline users in improving the quality of safety data entries.

On a wider scale, the last 15 months have seen AI thrust onto centre stage. Following huge waves of funding, AI research hubs are progressing at an alarming rate, the pinnacle of which is achieving real-world interaction artificial general intelligence (AGI). There is much debate as to what constitutes AGI in theory, but Verdantix summarizes it as an AI model that is able to solve complex problems across different modalities – text, visual and audio – well beyond the problems it encountered during its creation. This development, coupled with tech firms publishing algorithms as open-source, makes applying AI capabilities to novel use cases an accessible option. The most prominent AI developments recently have centred on generative AI (GenAI), driven by API-accessed text-generating large language models (LLMs) such as GPT-4 and image-generating models such as DALL·E 3, alongside open-source LLMs such as Llama 2 and Mistral-7B. All of the above models can generate new data based on user input text.

For the EHS tech market, increasing accessibility to market-leading open-source algorithms has the potential to blow the AI arms race wide open, offering both opportunities and risks for leading vendors. Early movers have already demonstrated applications for GenAI's ability to retrieve relevant data to improve capabilities and user experience. Take Avetta’s AskAva AI Risk Assistant solution, planned to be released to general customers in Q2 2024. The tool, built on an AI engine that combines Avetta’s proprietary data with ChatGPT, provides hiring clients and contractors with tailored safety content to manage risk.

Despite increasing accessibility to GenAI capabilities, don’t expect revolutionary EHS applications to flood the market. The use cases for open-source AI are still limited, and performance is heavily impacted by the quality of input data. Even using open sources, vendors will have to invest, time, money and effort in intense development and testing to produce a data orchestration solution that enables accurate, useful and trustworthy AI-powered tools that are ready for market. However, as open-source AI algorithms continue to improve, EHS software vendors should be abundantly clear: they will soon be at a competitive disadvantage if they’re not in a position to seize the opportunities of AI.

To access more research on EHS use cases for AI, please read the recent Verdantix report Strategic Focus: AI And the Revolution Of EHS Compliance. To explore more Verdantix EHS research, visit our research portal.

Tom Brown

Senior Analyst

Tom is a Senior Analyst in the Verdantix EHS practice. His current research agenda focuses on a range of EHS topics, including high-risk safety controls, contractor management, environmental services and EHS digitization strategy. Prior to joining Verdantix, Tom achieved a Master’s in Chemical Engineering from the University of Nottingham.