Ep 321 – Molly Graham (Host, WorkLife Podcast)

1. If you want employees to embrace AI, stop giving them reasons to fear it

Companies are asking employees to experiment with AI while the broader narrative tells them that the same technology could eventually eliminate their jobs.

Molly compared it to introducing someone new to the team, asking an employee to train them, and then telling them that person might replace them six months later. You can’t expect curiosity and experimentation to thrive when the underlying emotion is fear.

“The future is going to be defined by the people that can lean into AI with curiosity and with excitement.”

For People leaders, that means being extremely intentional about how AI is discussed — especially around organizational changes. Don’t casually use AI as the explanation for workforce reductions if the actual reasons are broader business decisions.

Your job is to create enough clarity that employees can stop wondering what leadership secretly knows and start experimenting with what the technology can actually do.​

2. Your AI strategy needs a story — and you’ll need to repeat it 4,000 times

Employees want to understand how leadership thinks about AI, what role it should play inside the company, how jobs might evolve, and what experimentation looks like today.

“Everybody’s an adult. They just wanna understand what’s going on and how to think about it.”

That narrative can’t live in one kickoff presentation or an AI policy buried in Notion. Leaders have to reinforce it in all-hands meetings, emails, Slack messages, manager conversations, and day-to-day decisions.

Molly’s rule of thumb: communicate something so many times that you start to feel ridiculous.

Because when the environment is changing this quickly, repetition isn’t overcommunication. It’s how you create stability.

3. Psychological safety isn’t built through messaging. It’s built through what leaders do after something goes wrong.

If you want people experimenting with new tools, they need to know what happens when an experiment fails.

Molly pointed to Facebook’s engineering culture during her time there as an example. When something broke, the immediate questions weren’t focused on blame. They were focused on understanding what happened and designing the system so it wouldn’t happen again.

One leader even described an engineer openly admitting that their code had caused a massive outage — and viewing that admission as evidence that the culture was working.

“It’s what you do, not what you say.”

Look at the questions leaders ask during retrospectives. Watch what happens when someone admits a mistake. Pay attention to who gets celebrated, what gets rewarded, and whether leaders punish failed experiments after publicly encouraging people to take risks.

The culture employees experience in those moments will overpower anything written on the wall.

4. AI means manager training may need to become employee training

The rise of the “super IC” sounds like a future with fewer managers. Molly thinks the opposite might be true.

As more employees begin directing agents, everyone starts doing a version of management. The skills required to get useful work from AI — setting context, creating clarity, giving instructions, aligning work to goals, reviewing output — look remarkably similar to the skills required to manage humans well.

“Every single person in the world just became a manager.”

That creates a new challenge for People leaders.

Manager training can no longer be reserved only for employees who have direct reports. Companies may need to teach far more people how to set expectations, communicate priorities, provide useful feedback, and create alignment.

Because if an employee gives unclear instructions to one person, they create confusion. If they give unclear instructions to 100 agents, they create confusion at scale.

5. Nobody has AI figured out — so stop looking for the finished playbook

There’s enormous pressure right now to find the company that has cracked AI transformation and copy what they’re doing.

Molly’s message: that company probably doesn’t exist.

Even sophisticated organizations are switching models, changing workflows, testing tools, abandoning approaches, and trying again weeks later.

“I don’t think there’s anybody out there that’s like, ‘I got this. Let me show you how. Here’s the playbook.’ And if they do, the playbook expires like three weeks later.”

That should change how People teams approach AI transformation.

Instead of asking, “Who has solved this?” ask:

• What can we test this week?
• What did we learn?
• What should we stop doing?
• What previously failed experiment might be worth retrying now?

The advantage won’t come from discovering the permanent AI operating model first. It will come from building an organization that can learn faster than everyone else.​

6. AI can make senior employees act like interns — and leaders need to restore accountability

One of Molly’s biggest concerns is that experienced employees are sometimes using it without exercising the judgment that made them experienced in the first place.

Someone generates a polished document, barely reviews it, and forwards it along. A presentation looks sophisticated, so nobody stops to ask whether it’s actually correct, useful, or connected to the conversations that led to it.

Molly’s framing: treat AI like an extremely confident intern.

“I don’t care how you made it, you own it.”

If an intern handed you a presentation, you wouldn’t automatically forward it to the CEO. You’d review it, challenge the thinking, correct mistakes, and probably send it back for another round.

The same standard should apply to AI-generated work.

And the more senior the leader, the more important this becomes. Executives who ship unreviewed AI output aren’t just producing bad work — they’re signaling to the entire organization that accountability is optional.

‍