Prads, I know you are real, but have you outsourced your post composition to a chatbot, by any chance?

Morning Prads

Getting back to the real world, there have been many threads on these Forums about trying to use proprietary systems to ease the job of the OSH professional.

Some have offered ideas as to suitable providers.

Others have said that for most of what you need you could ask some techie colleague to put together a spreadsheet or whatever to give you the information you need or to help with things like interactive form filling.

The problem is that the proprietary tools tend to be fine for simple processes that don't change much but are not so adaptable as HUMANS when it comes to thinking about what is different between two scenarios which at first glance look very similar.

....and AI is only as good as the way it has been designed by HUMANS who might not be the right HUMANS for the problem you come across tomorrow, as it hasn't been programmed into the way the AI thinks.

Finally, the AI sits in an office (in hyperspace) and is not there at the coal face, except when using a drone which operates at some distance from the coal face and can't ask the worker WHY they are not doing this, that or the other.

Hence, the drone is very good at spotting that the worker isn't wearing some item of mandatory PPE but not so clever when it comes to considering whether the mandatory rule is sensible, let alone thinking about how the conditions could be improved so as to negate the need for space suits.

....and of course if the AI comes across a situation where there are two risks co-existing and the authoritative guidance for one risk says "Do X" and the authoritative guidance from another source that hasn't taken account of the other risk says "Do Y", then AI is bamboozled as it can't think about whether Doing X and Y together could either negate the efficacy of X or Y, or change the risk profile to exacerbate the problem. 

An unregulated piece of software with full access to the internet and all of its false info, misconceptions, and sales slants on what is required, not to mention snake oil touters. With no sense of empathy, feelings, or morals. Yeh let it make EHS decisions.

What could possibly go wrong.

I can’t think of any area where AI would be useful, even data retrieval. Lots of info is not in the public domain from HSE anymore and you have to pay the specialist body. So don’t let the AI know your credit card number.

Where was the original poster thinking it could be used?

That's a fair point achrn, there may well be a role for AI in trawling existing sources, but the value added bit is likely to be where the practitioner earns their keep.

I see it as a tool a bit like say a RAMS software package, but not as revolutionary as the tech boys are claiming, because ultimately we are all about goal setting/ continual improvement and not repeating what has come before.