Comments on: ChatGPT article on psychosocial hazards at work

By: skillr01

skillr01 — Mon, 06 Feb 2023 02:50:56 +0000

In reply to Kevin Jones. Kevin, yes, but an interesting perspective nevertheless.

By: skillr01

skillr01 — Mon, 06 Feb 2023 02:50:09 +0000

In reply to Fred Goodwin. Greta observations Fred

By: andrew.hopkins@anu.edu.au

andrew.hopkins@anu.edu.au — Wed, 01 Feb 2023 11:03:16 +0000

Thanks for exploring this issue. Have you seen this?
https://theconversation.com/major-publishers-are-banning-chatgpt-from-being-listed-as-an-academic-author-whats-the-big-deal-198765?utm_medium=email&utm_campaign=Latest%20from%20The%20Conversation%20for%20February%201%202023%20-%202531325426&utm_content=Latest%20from%20The%20Conversation%20for%20February%201%202023%20-%202531325426+CID_02fc927c904c025f08dc6c7b5d620a1d&utm_source=campaign_monitor&utm_term=Major%20publishers%20are%20banning%20ChatGPT%20from%20being%20listed%20as%20an%20academic%20author%20Whats%20the%20big%20deal

By: Faith Eeson

Faith Eeson — Fri, 27 Jan 2023 08:48:45 +0000

Fred I agreed. I too have been testing ChatGPT and it provides very basic information. You as the user needed to be knowledgeable on the subject matter, ask the questions and then fill in the gaps.

By: Mick Peterson

Mick Peterson — Thu, 26 Jan 2023 23:23:45 +0000

Yes misses a lot of organisational causes. Safe person approach which has a place but seriously limited from a preventative point of view.

By: Kevin Jones

Kevin Jones — Thu, 26 Jan 2023 22:04:10 +0000

In reply to Graham Dent. I suspect this software will replace the need for humans to produce a “rough draft”. But it creates more importance for knowledgeable editors and fact-checking.

By: Graham Dent

Graham Dent — Thu, 26 Jan 2023 21:49:24 +0000

I am curious. If the current WHS regulations, codes and guidance on psychosocial health were uploaded and it was asked to write a new (and better) one, how would it fare?

I wonder if Regulators will attempt to do this in the future as a starting point on developing guidance in new areas.

For example uploading relevant accident descriptions (as set out by courts, coronial inquests, regulator summaries, statistical data, overseas models), and asking the AI to write a draft or a consultation paper.

Obviously it would require detailed human review, but l suspect it will be something which someone will experiment with sooner rather than later.

By: Fred Goodwin

Fred Goodwin — Thu, 26 Jan 2023 21:13:49 +0000

I've been using ChatGPT a bit testing it's capabilities. Given you have asked it a pretty broad question it has given you a pretty generic answer. If you run it again asking it to factor in the bits you thought were lacking the answer will improve. We asked it a generic safety question and felt the same. Went back and refined the question a few times and the answer was unsettlingly accurate and knowledgeable. It's "learning" stopped about 2 years ago so anything current won't be there for it to use for reference, but there is talk the pro version will connect live to the internet. One area we are thinking will be interesting is if we can load all the organisation WHS policies, procedures, and relevant legislation then ask it how do I questions, or even to look for inconsistenties across the documents. Upload isn't working at the moment, but we cut and pasted some documents and it was (again creepily) good at churning through the documents. Will it replace advisors, or even blog writers , probably not in the near future, but it does have the potential to make life easier.