Worthwhile Content: February 2026
Some worthwhile reads, watches, and listens from the last month
World of Work:
In February, I discussed the delegation in two parts. First, was an overview of the two reasons to delegate work and then part two explored a framework I devised called the ‘Delegation Contract’, which is designed to be used every time you delegate work. I responded to a reader question about a boss ignoring an email about a promotion. Finally, I wrote about how AI won’t take as many jobs as the technology might theoretically allow, as adoption will lag behind capability in most organizations. That’s a lot for a short month! Some other worthwhile content I digested in February include:
This HBR article gives some decent advice to senior leaders on how to brag about your success without, well, bragging about your success (paywall).
On the government side, OPM says its biggest challenge is “creating a high-performance culture” across the federal government—but the critique here is that the current push is mostly nuts-and-bolts (ratings, bonuses, consolidating HR data) without tackling the harder stuff: leadership, behavior change, psychological safety, and capability-building. The argument is that government won’t get better results just by tweaking HR mechanics; it needs to link people practices to agency performance and rebuild trust after staffing cuts and morale hits.
Glassdoor tracked 268K people who left two employer reviews and found a simple pattern: the worse the first company rating, the more likely you are to leave (shocker)—and leaving often pays off! People who rated their job 1–2 stars were far more likely to see their next rating jump to 3+ stars if they switched employers, especially if they landed at a “Best Places to Work” company.
Bloomberg reports that as Wall Street pulls back on DEI, some Black bankers at JPMorgan and Citi say the day-to-day experience is getting tougher. Less support, more uncertainty about how to grow, and a lingering sense that the rules are shifting mid-game. More than two dozen Black and other minority bankers described the rollback as frustrating and confusing, especially as career-making opportunities often depend on informal sponsorship and visibility (paywall).
AI & Work
What do people do when the job market isn’t great? They go to law school! But the payoff is less predictable now, thanks to tighter student lending rules and A.I. that’s already changing how legal research and document review get done.
The authors of this study build AI personas of real CEOs using their own communications, then test whether those personas make decisions in believable, measurable ways. The results show these theoretically guided personas can closely mirror human moral reasoning on key scenarios.
I’m a big fan of voice dictation. I take a lot of voice notes and use voice dictation for AI all the time. Bloomberg reports that more people are using this approach to draft long messages and give detailed instructions to chatbots, not just dash off quick notes… in crowded offices. Whether it sticks may depend less on the tech and more on whether workplaces can handle the awkwardness (and noise) of everyone talking at once.
A report cited by The Information says OpenAI has created an internal version of ChatGPT to help investigate leaks by comparing what shows up in press coverage with what existed inside company channels (like files, Slack threads, or other internal records). Is the use of ChatGPT for employee surveillance by its creators a sign that it will soon be ubiquitous, or just a sign that only the most AI-native organizations are ready for this approach?
Amy Farner at Bersin argues that as AI takes over more task orchestration, managers will spend far more time coaching, developing, and supporting employees (and even managing AI agents), with “HR-style” responsibilities potentially jumping from ~10% of the job to the bulk of it. Companies may have to rethink who should be a manager in the first place, because people skills become the main requirement.
The reason GenAI is scary for employees might be because it distinctly threatens the core psychological needs of competence, autonomy, and relatedness.
My friend Nick Thompson shared a walk-through video that explains how to stop the various GenAI tools from training on your data. Gemini is the hardest to shut off.
AI wiz Professor Ethan Mollick gives a great primer on which AI tools to use in the current agentic era.
In this blunt anti-hype piece, David Williams Silva tries to pop the “AI cult” bubble by separating what LLMs do well (fast pattern-matching and language generation) from what they don’t (true understanding, common sense, causal reasoning). Silva praises Yann LeCun’s skepticism about LLMs as a path to human-level intelligence, and criticizes the incentives that reward dramatic claims from founders and celebrity researchers. The message is less “AI is useless” and more “the hype is the real scam.”
General Interest
I don’t want to give my children a smartphone until they’re 35. But I definitely won’t give them one before teenage years. One analysis of 10,500+ children found that getting a phone at age 12 (vs. 13) was linked to notably higher risks of poor sleep and obesity, while other research connects high or “addictive” use to attention and learning dips, as well as higher risk of suicidal thoughts in vulnerable groups (paywall).
This is just insane. After a family kayaking and paddleboarding trip turned dangerous, 13-year-old Austin Appelbee made a “superhuman” decision: ditch the kayak and even his life jacket and swim for help. He made it to shore after roughly four hours, called emergency services, and a multi-agency search eventually rescued his mom, brother, and sister that night.
Studying musicians while they actually perform has been a nightmare for neuroscience — you can’t put a pianist in an fMRI and you can’t get perfectly repeatable performances for EEG. Pianist Nicolas Namoradze helped crack that problem by “finger-syncing” to a high-precision player piano version of his own playing, allowing researchers to isolate performance-related brain signals. Now scientists can ask more realistic questions about how planning, movement, memory, and emotion light up during live music-making. Super cool!
This story is a wild ride. A whistleblower inside a Laos-based scam factory (“Red Bull”) secretly documented how workers create fake profiles (including AI deepfakes) to run romance-and-crypto cons on victims worldwide, then tried to flee. When the escape went wrong, he was assaulted and threatened, but still managed to get out and deliver thousands of pages of internal evidence to researchers and journalists. Incredible work from Wired (paywall).
Is music streaming about to break? The model doesn’t support artists who want to connect with fans. I’m not sure I agree, but it’s a good take.
Musings
I’m over February. Over it. Give me warm May weather.