Move Fast ... and then fix things.

ChatGPT is probably about as wrong as a human is ... so we can incorporate it into our processes the same way we do humans. But boy can it write clickbait!

• Leadership Moments: Chatting with GPT "Uncover the Truth Behind ChatGPT: Is this AI Model More Hype Than Reality? Find Out in Our Exclusive Leadership Moments Feature!"

• One Minute Pro Tip: No Surprises "Transform Your Performance Reviews in 1 Minute with this Pro Tip on No Surprises!"

• Chapter Teaser: The Red Ink Rule "Ready to Skyrocket Your Productivity? The Red Ink Rule Will Change the Way You Work Forever!"

• Appearances "Don't Miss Out on These Must-See Appearances by a Leading Expert in Tech and Security!"

• (paid) Essay: Decision Making "Unlock the Secrets of Successful Decision Making with a Behind-the-Scenes Look at the First Chapter of a Bestselling Book!"

New section descriptions written by ChatGPT.

Leadership Moments: Chatting with GPT

ChatGPT is a Large Language Model (LLM); a form of AI trained on a very large set of written language. What makes ChatGPT fascinating is that, unlike early versions of language AI, entire paragraphs and essays produced by ChatGPT make sense. ChatGPT has mastered the art of fluently sounding like a human, and, backed by a significant body of factual data, has the ability to also be correct quite often.

Unfortunately, quite often isn’t always often enough. ChatGPT is that person at a party who can confidently expound seemingly endlessly on topics far and wide, but you’re never quite sure when your leg is being pulled, and when you’re learning new (and true) information. ChatGPT doesn’t necessarily grasp the nuance underlying certain word phrasings and may lack context that enables (some) humans to navigate dangerous interactions.

It’s easy to fall into an extreme camp: either ChatGPT is completely transformative, or it is completely worthless. The reality is somewhere in between. Writing this essay took me about 10 minutes; ChatGPT generates a similar one (but not as useful, in my opinion) in about 30 seconds*. That’s a helpful aid to anyone doing content generation, because it’s often faster to edit than it is to write coherent first draft. That’s evolutionary, but not revolutionary. Overly strong reactions for or against ChatGPT may keep you from using it appropriately.

*After this post goes live, I’ll post the ChatGPT prompt/response as a comment.

One Minute Pro Tip: No Surprises

Many organizations have an annual performance review cycle, where once a year a manager documents for an HR file how an employee is doing. In theory, it’s an opportunity to help develop an employee and direct their course for the year; in practice, it provides documentation for the company about the excellence (or lack thereof) of an employee in the event of future legal action.

If you’re the manager, you have one job: don’t surprise your employee. You have the entire rest of the year to tell an employee what they’re doing well, and what they need to improve on. This is not the day to spring on them that you’re unsatisfied with their performance. If you have been telling them how wonderful they are in person, and then you write down all their faults for the first time? That isn’t helping develop them in any way at all. Tell them their faults in person, and write down how wonderful they are for HR.

Chapter Teaser: The Red Ink Rule

Ch 37: It’s easier to prompt a reaction than an initial action.

Asking people – or yourself – to do work on a blank slate is hard. They might not know how to start. They might be worried that they don't really know what you want. Picking up a black (or blue!) pen for the first time is daunting. But everyone loves picking up a red pen. Put inferior quality work in front of someone, and they'll quickly dive in to making it higher quality.

This can work on yourself — write anything (or have ChatGPT write it for you), and then dig in and edit. Working with a partner, tell them you’re providing a bad draft, and that you welcome edits. Maybe you just provide an outline, and then invite digging in.

Sometimes, a bad sketch is the quickest way to get started.

Appearances

• Feb 9, 2023: Chatted with the TechExecs Network about Budgeting for Cloud Security, although we mostly talked about budgeting and uncertainty. (video)

• Feb 9, 2023: Recap episode of Cloud Security Reinvented on Board Buy-In, Vulnerability Management, and Policies.

• This week: I’ll be at dinners for Valence Security & Grip Security in the Boston area. If you’re a security leader interested in attending, ping me and I’ll connect you to the hosts.

Behind the paywall: the intro chapter for the first book I started to write, on decision making…