It depends upon the context. These AIs are trained using information available on the internet so for the average person asking about common health and safety issues, ChatGPT will generally give good advice. For specific technical issues and bespoke advice, I would be more careful.

I asked ChapGPT "How dangerous is handling paper in an office"

And got back an answer that will keep people busy for ages

"Handling paper in an office is generally not considered dangerous, but it can still present a few risks if proper precautions aren't taken. Here are some potential hazards:

Paper Cuts: One of the most common and minor injuries that can happen while handling paper is paper cuts. These small, but painful cuts usually occur when handling sharp edges of paper.

Slips, Trips, and Falls: A pile of paper on the floor or paper scattered around the office can create tripping hazards. This can lead to slips or falls, especially if the office is not tidy.

Ergonomics: Repeatedly handling and organizing paper can lead to poor posture or strain. For example, bending over desks or reaching for files can cause back, neck, or shoulder pain. This is especially true if the office is not set up ergonomically.

Air Quality: A lot of paper in an office can contribute to dust and particles that can affect the air quality. Prolonged exposure to dust or paper particles may aggravate allergies or respiratory issues, especially in poorly ventilated areas.

Fire Hazard: Excessive amounts of paper stored improperly can pose a fire risk. If paper accumulates too much, it can catch fire easily, especially near heat sources or electrical equipment.

Mental Stress: An overabundance of paperwork can also contribute to mental stress or anxiety. This can affect productivity and employee well-being, as they may feel overwhelmed by the clutter.

To minimize these risks, it’s important to:

Keep workspaces organized and free of clutter.
Take regular breaks to avoid strain.
Use ergonomic office furniture.
Ensure good ventilation to reduce dust.
Store paper safely away from potential fire hazards.

Gotta use it, it's the future (and the present.) The H&S profession need to take advantage of it. It can only be good. It's very helpful, efficient and saves time across a multitude of areas. Not too much to debate in my view. But use the Trust and Verify principle.

Very useful for ISO audits.

Example:"Not having a record of the incident" Which clause of the ISO45001 does this not comply with?

The clause of ISO 45001 that "not having a record of the incident" would likely not comply with is Clause 9.1.3, Monitoring, measurement, analysis, and performance evaluation.

Here's why:

Clause 9.1.3 requires an organization to "evaluate the performance of the OH&S management system," which includes monitoring and measuring relevant data. Incident records are crucial data for performance evaluation.
Missing incident records indicate a failure to monitor and measure incidents, which is a key aspect of a safety management system. Incident records provide data on the frequency and types of incidents, helping organizations identify trends, assess the effectiveness of their controls, and implement corrective actions to prevent recurrence.
Additionally, incident records are often linked to other clauses of ISO 45001, such as:
Clause 8.1.3, Management of change, where incidents can trigger a review of existing controls and necessitate changes.
Clause 10.2, Nonconformity and corrective action, where incidents represent nonconformities that need to be investigated, and corrective actions need to be taken to prevent recurrence.
Clause 7.5, Documented information, as incident records themselves are a type of documented information.
Therefore, a lack of incident records represents a clear deficiency in monitoring, measurement, and performance evaluation, violating Clause 9.1.3.

It depends on the framework. Definitely for audits and for research purposes. Also exploring the use of ChatGPT to create workflows for efficiency. But also need to be aware of information overload. Strong believer in automation and getting away from paper based processes. Journey of learning, trail and error to find the right fit

Didn't someone here comment a few months back that they had been reviewing an NZ based clients hazard register, and figured it was AI generated because it listed snake bites as a likely hazard with outdoor work?

I am leaning towards a strong level of distrust because the average user won't frame the question with enough context to get a useful answer, and will blindly trust the answer.