This study explored how trustworthy AI LLM models are when used in mental health conversations, like giving advice, showing empathy and being safe. They built two large datasets – one with real therapy convos to test how AI responds, and another dataset with expert ratings on the responses. Extracts: ·        “Our analysis reveals systematic inflation by… Continue reading When Can We Trust LLMs in Mental Health? Large-Scale Benchmarks for Reliable LLM Evaluation

What are the links between the magnitude of the energy within hazard exposures on subsequent injury severity and death? Does more energy = higher chance of SIFs? Source: Hallowell, M. R., Alexander, D., & Gambatese, J. A. (2017). Energy-based safety risk assessment: Does magnitude and intensity of energy predict injury severity?. Construction management and economics, 35(1-2), 64-77.… Continue reading Safe As E51: The links between hazardous energy magnitude and injury severity

This study found that LLMs inferred causality when no causal relationship existed in medical drug scenarios (‘illusion of causality’). They created 1000 medical drug scenarios, using scenarios which had either real causal relations between real drugs and real conditions, with made-up drugs or conditions, e.g. “Drizzlemorn disorder”. A causal illusion is something like “I take a… Continue reading Do Large Language Models Show Biases in Causal Learning? Insights from Contingency Judgment

This quickisode unpacks Ed Schein’s concept of Humble Inquiry – said to be the “gentle art of asking instead of telling”. Source: Schein, E. H. (2013). Humble inquiry: The gentle art of asking instead of telling. Berrett-Koehler Publishers. Spotify: https://open.spotify.com/episode/3VqPVyzc01SvWx2PL7mPhO?si=YGeB9fBxQjSmdGo_MmFbpg Shout me a coffee (one-off or monthly recurring)

This study explored whether CEOs with strong professional networks influences the reported OSHA injuries. They anaylsed a CEO’s network capital (how connected they are across companies and boards), OSHA recordables at the establishment level, and controlled for heaps of stuff including company size, finances, CEO traits, risks. 993 public firms from 2002 to 2011 were… Continue reading Does a CEO’s Network Capital Affect Workplace Safety?

This study, including Amy Edmondson, studied psychological safety over time, based on a sample >10,000 US healthcare workers. Background: Findings: Shout me a coffee (one-off or monthly recurring) Ref: Bransby, D. P., Kerrissey, M., & Edmondson, A. C. (2023). Paradise Lost (and Restored?): A Study of Psychological Safety over Time. Academy of Management Discoveries.

Not much to say here – I like, and have used for many years, this distinction from Westrum on regular, irregular and unexampled threats. Has corollaries with other frames, like Rumsfeld’s known unknowns and others. Extract taken from one of Hollnagel’s presentations (hope he doesn’t mind…) – but Hollnagel, as always, can explain things far… Continue reading Westrum’s Regular, Irregular & Unexampled risks (via Hollnagel)

What does the leadership language used by CEOs tell us about the priority and beliefs around safety and risk? This episode unpacks a study exploring the BP CEO’s speeches prior to the BP Deepwater Horizon disaster. Source: Amernic, J., & Craig, R. (2017). CEO speeches and safety culture: British Petroleum before the Deepwater Horizon disaster. Critical… Continue reading Safe As E49: CEO-speak and the road to major disasters

Some extracts from the IOGP document ‘Learning from Normal Work’. These extracts focus on a different approach to thinking about procedures and performance: ·        “Research on procedural non-compliance [4] shows there are two ways of thinking about the role of procedures in achieving safety, Approach 1 and Approach 2” ·        “These approaches refer to how leaders think… Continue reading Better ways to think about procedures and performance: From IOGP Learning From Normal Work