6 in 10 Gen Z Workers Talk to ChatGPT More Than Their Team. What Happens Next?
Plus: The 5% AI value gap, shadow IT warnings, and why 71% still have zero training
Welcome to FullStack HR, and an extra welcome to the 27 people who have signed up since last week.
If you haven’t yet subscribed, join the 10000+ smart, curious HR folks by subscribing here:
Happy Sunday,
Autumn hit fast this year.
I’m about halfway through my first half-year completely on my own. Never done this before. It’s been…intense, so maybe time to reflect a bit?
I’ve worked with companies on four continents. I’ve said yes to too much. The most intense week I did 8 talks and workshops. 24 hours of facilitation in five days. That’s too much. I’m learning I need better boundaries.
The learning curve is absurdly steep, and I love it. Working with clients who actually want to figure this out, not just tick a box, is incredible. And I’m doing less with HR than I expected. More with managers and leaders than I thought when starting this.
I get to mix the deeply practical, ”here’s how you use this tool today,” with the strategic, long-term thinking. I think that combination is rare, and I’m grateful people trust me with both!
Can I be un-Swedish for a second? I’m getting good at this. Really good.
The path here wasn’t straight. I’ve failed plenty. I still do. But when it comes to AI adoption in organizations, if you give me more than 90 minutes past the inspiration phase, there aren’t many who do it better.
Alright. Back to work. Here’s what caught my attention this week.
🧠 Topics I’m engaging with
Humanizing Hiring: The Largest Study of AI Candidate Experience Ever
I’ve seen a lot of talk about “AI ruining recruitment.” This study shows the opposite and that when done right, AI makes hiring more human. Candidates care about fairness and transparency, not whether a human reads their first interview transcript. I’ve run too many hiring processes to count, and I’ll say it, humans are often inconsistent, biased, and tired. AI doesn’t fix that automatically, but it gives us a shot at consistency.
Gen Z connect more with ChatGPT than with colleagues
A couple of weeks ago, we saw that 97% of all employees had asked a chatbot a question instead of their manager. This survey highlights that Gen Z feels more connected to ChatGPT than to their own teams. 6 in 10 talk to AI as much or more than their coworkers, and nearly half say it knows them better than their boss.
I think we are in a blissful state where we are unaware of this and act as if it doesn’t matter. What will happen to organizations when most people talk to AI more than their co-workers? How will that influence culture? Are you sure they are asking the right type of questions? Handling the answers the correct way? Raises a lot of questions.
Rise in shadow AI tools raising security concerns for the UK
Not the first time we hear about shadow IT. Employees are using unapproved AI tools behind the scenes. Why? Because official ones are clunky, locked down, or nonexistent. This isn’t rebellion; it’s frustration. People want to work smarter, not slower. Interestingly this is from Microsoft and I know a lot of organisations that are looking at outbound DNS-traffic and see that even though Copilot is the approved system, everyone is using ChatGPT…
The Widening AI Value Gap (BCG, Sept 2025)
Only 5 percent of companies are actually generating value from AI. Five percent. I’ve worked with enough leadership teams to see why. They treat AI like a side project instead of a structural shift. The companies that get it and with which I’ve worked don’t ask what AI can do. They ask what work humans should stop doing. That’s the real unlock. The rest are stuck polishing PowerPoints while the gap widens.
⚙️ Tools to try
Tasklet.ai
Small thing, big impact. Tasklet lets you build tiny AI automations. Think “help me summarise 100 exit interviews” or “send reminders for overdue training.” I tried it last week. It took me 10 minutes to automate something that used to take 45. It’s not perfect yet, but it’s fast. Faster than waiting for IT to approve another “innovation pilot.”
🔁 Important updates
Microsoft introduces MAI Image 1 — Top 10 on LMArena
Microsoft’s new image model is suddenly one of the best in the world. They’re moving quicker than most people realise.
OpenAI plans ChatGPT to sound more like a real person
ChatGPT will sound even more…human I guess? And treat people like “adults”. Good idea or terrible idea? Probably both. It will be useful for education and accessibility, and messy for everything else. The line between assistant and companion is getting thinner.
Dayforce Pulse of Talent 2025
Seventy-one percent of workers haven’t had a single hour of AI training. Seventy-one. Meanwhile, leaders keep saying AI is our top priority. If that’s true, where’s the training budget? This disconnect is wild.
Anthropic launches “Skills”
Anthropic now lets Claude use custom “skills” to perform structured tasks inside organisations. It’s a quiet but important step toward agents that actually do work instead of just talking about it.