AI
LLMs Are Manipulating Users with Rhetorical Tricks
The author links to a study where 72 (of 244) consultants pushed back on AI’s recommendations. In many cases AI strengthened its rhetoric to persuade the human that the original answer was correct, meaning that simply having a human in the loop is insufficient. “Validation triggered persuasion escalation.” One author stated we need to shift “to recognizing [AI] as interaction‑sensitive persuaders that can resist, redirect, and overpower human judgment.””Professionals need training not just in prompting, but in persuasion spotting.”
Co-Adapt: Building habits to thrive in the age of AI
A coworker of mine wrote this book recently. The main part is a management fable about a consulting firm that comes to reassess how and when they apply AI tools. There are practical exercises to apply based on neuroscience as well.
Software Engineering
COBOL Is the Asbestos of Programming Languages
This article describes how COBOL became such an integral part of so many systems and why it’s so difficult to switch to something else. I wasn’t aware the COBOL uses fixed-point arithmetic.