Read Seaways online at www.nautinst.org/seaways July 2023 | Seaways | 17 Providing learning through confidential reports – an international co-operative scheme for improving safety MARS – Lessons Learned MARS Report No 369 July 2023 Visit www.nautinst.org/MARS for online database To conclude, I would humbly submit that by reading MARS reports, and by contributing your own accidents and incident reports to MARS, we can lend a hand at the fourth element listed by Chat GPT-4, that of Lessons Learned. Paul Drouin FNI, Editor, MARS MARS 202330 Dropped object incident causes minor injury  While disconnecting the cargo hose on a tanker with the ship’s crane, a member of the shore personnel working on deck suffered a minor injury from a small piece of wood that fell from height. The wooden piece was sheathing placed around the boom’s ’parking crutch’ to avoid metal-to-metal contact. Over time, the securing arrangement for the sheathing had come loose, and when the boom was lifted, it slipped from position and fell, striking the shore crew. The victim was given first aid and was in good medical condition thereafter. The company investigation found, among other things, that the monthly maintenance of the crane was not carried out as per the Planned Maintenance System (PMS). Visual inspection of the sheathing would likely have identified the wear and probably avoided the accident. Lessons learned l Monthly PMS inspections are established for a reason and should be carried out with due diligence and ‘fresh eyes’ at each repetition. l In this instance, no specific PMS task was assigned to checking the condition of the wood sheathing. When new hazards are discovered, risks can be reduced by creating a specific PMS task for the object of the hazard. n Editor’s Note: With all the recent press coverage about the power of Chat GPT-4, I decided to run my own specific test, a question about learning from accidents. While the large-scale multimodal model on which GPT-4 generates its output is far from infallible (GPT-4 has been known to ‘hallucinate’ facts and make reasoning errors), the answer to the question came back in about two seconds and, in my opinion, was quite good. (March 14 version of Chat GPT-4) Q: Why is it that even after in-depth investigations into accidents that reveal the contributing factors, the same mistakes are made and similar accidents happen again? GPT-4 Answer: There are several reasons why the same mistakes are made and similar accidents happen again, even after in-depth investigations into accidents have revealed the contributing factors. Here are some possible explanations: Human Error: Humans are fallible, and even with the best training, people can make mistakes. While investigations can identify contributing factors and provide recommendations for improvement, it is ultimately up to individuals to implement these recommendations and make changes to their behaviour. Unfortunately, even with the best intentions, people can slip back into old habits and repeat the same mistakes. Organisational Culture: Organisational culture plays a significant role in safety. If an organisation has a culture that prioritises production or efficiency over safety, then accidents are more likely to occur. Investigations may identify this culture as a contributing factor, but changing an organisation’s culture is a long-term and challenging process. Complexity: Many accidents occur in complex systems, such as transportation or healthcare. These s