A couple of days ago I posted an analysis of three recent BSEE Safety Alerts. The obvious insight was that Mechanical Integrity was the SEMS element most at play in these incidents. What was less obvious is that luck was also a factor ― in one case there were minor injuries, but in each case someone could have been killed.
ChatGPT’s Analysis
I wondered what ChatGPT would make of this situation, so I asked it the following question.
BSEE (the Bureau of Safety and Environmental Enforcement) recently issued three Safety Alerts: 493, 494 and 495. Which elements of SEMS failed in each of these incidents? What other factors would you consider to be important when analyzing these reports?
The response was good. First, it issued a sensible disclaimer, ‘While the specific details of these incidents are not fully provided in the available sources . . .’ It then correctly identified Mechanical Integrity as the most important factor in each case. It also identified other SEMS elements, such as Operating Procedures for the handling of nitrogen cylinders.
It did not call out luck ― which is obviously a subjective topic. So, I entered the following supplemental question, ‘Was luck also a factor?’
Once more, it gave a sensible reply, ending with the following statement.
‘. . . relying on luck is a dangerous approach. A robust Safety and Environmental Management System (SEMS) is designed to reduce uncertainty and eliminate the need for luck by proactively managing risks.’
Timeliness
I concluded this thought experiment by asking ChatGPT,
Are you familiar with the post from Ian Sutton that was published a few days ago? Its title is ‘Three BSEE Safety Alerts’.
It replied.
I couldn't locate a post titled "Three BSEE Safety Alerts" by Ian Sutton from a few days ago. However, Ian Sutton has previously discussed BSEE Safety Alerts, such as Safety Alert No. 459, which addresses the critical importance of proper training in preventing fatalities.
The post to do with Alert #459 was published on December 23rd 2023.
Why Bother?
ChatGPT gave a solid response to do with these three Safety Alerts. Which begs the question, ‘Do we even need human process safety experts any more?’
If the program can generate high quality responses, then clients will not bother to hire those experts, and the experts will not bother to do the hard work of analyzing information. After all, why should they give their expertise to ChatGPT for free? The upshot could be that the experts will stop publishing their thoughts on public forums.
This being the case, ChatGPT will likely end up eating its own tail, coming up with ever more absurd results.
Agree Property damage no luck. Luck was a factor in the outcome of injuries and fatalities.