“Our profession's ethics canons — duty to workers, scientific integrity, transparency, independent professional judgment — get harder, not easier, when AI is in the loop,” says Mark Katchen, MS, MBA, CIH, founder and CEO of The Phylmar Group consultancy, in an email to ISHN.

Mark is one of the foremost ethics experts in the EHS field. He has recently co-written “Introducing Ethical Decision-making Occupational and Environmental Health and Safety: A Comparative Study Approach,” taught university courses on ethics and has discussed professional ethics on CNN, online workshops and numerous articles.

He sees artificial intelligence challenging EHS ethics in these ways:

Worker privacy and consent

“Many AI safety tools learn from injury reports, near-miss data, sensor readings, even biometric monitoring,” Mark says.” Workers have a stake in how that data is collected, used, and aggregated. Are practitioners disclosing when AI is part of an assessment? Should we be?”

Bias and equity

“Risk-assessment models trained on incomplete or skewed data can systematically under-protect specific worker populations — temporary workers, language minorities, contractors. Without scrutiny, ‘objective’ AI can launder bias.”

Accountability 

“What happens when AI gets it wrong. If an AI-recommended control fails and a worker is hurt, who's responsible — the practitioner who relied on it, the vendor who built it, the employer? Right now, this is murky, and ambiguity tends to favor the parties with the most resources.

Substitution of professional judgment

“When does using AI tip from ‘augmenting my expertise’ to ‘substituting for it’? The line matters — and our canons of practice don't yet have clean answers.

More broadly, Mark sees increasing use of AI creating two issues for the EHS field:
 
Accountability infrastructure

“Specific to our profession this includes CIH sign-offs, PE stamps, expert testimony, OSHA-recognized qualifications. AI doesn't carry professional liability, can't be deposed, and can't be the responsible party on a regulatory submission. That puts a structural floor under headcount in regulated workplaces, regardless of how capable the tools get.”

Task vs. Job

“The workforce research I find most useful treats AI as a task automator, not a job replacer — for almost every occupation only a fraction of the underlying tasks are fully automatable, and the rest get reshaped or augmented. The more interesting question isn't whether jobs disappear, but which tasks AI absorbs, what new ones emerge, and how the role itself evolves.”

 “These questions and issues aren't theoretical,” says Mark. “They're showing up in practice now, and they're exactly the conversations I think ISHN readers need to be having.”