HR on AI: What does all of 'this' mean for business?

The Workplace Intelligence newsletter take a closer look.

Note from the editor: The Workplace Intelligence Newsletter is a must read. It's weekly and it is packed full of valuable HR business insights. This week they tapped into some of their HR experts to really break down the impact of AI. You can read the whole thing in their newsletter, however, it's long...so to ensure you see some of the major takeaways, I used Claude to summarize for you. Here are 7 key takeaways.

1. Headcount replacement is the overhyped story

Boards keep asking which roles AI will eliminate. So far, the data says that's mostly not happening. McKinsey found that 89% of companies use AI somewhere, but only 37% see any measurable impact on operating profit. Just 14% actually cut headcount last year, against the 32% who predicted they would.

Leaders also think AI is doing more than it is. Atlanta Fed research found CFOs perceive about three times the productivity gain their own revenue and headcount numbers actually show.

AI is still paying off. The payoff is just showing up somewhere other than the org chart.

2. The bottleneck moved from making work to checking it

AI made first drafts cheap. Someone still has to review them, and that takes time. Workday found roughly 40% of the time AI saves gets handed right back as rework.

Our sense of the gains can be off, too. In one METR study, experienced software developers using AI were 19% slower, while believing they were 20% faster.

The technology is rarely what slows a company down anymore. The holdup is who has the authority to act on AI output, and whether managers trust it enough to stop re-checking every line. Companies that sort out those decision rights will move faster than companies that keep buying more tools.

3. The work that's left is harder

AI handles the easy, repetitive tasks first. That means everything that still reaches a person is tougher. It takes more judgment, more focus and more emotional energy. Productivity goes up while the employee's day gets harder.

In one study, 83% of employees said their workload went up with AI, not down, with burnout hitting hardest at the entry level. When you can always do a little more, you usually do.

This sets up a collision. Employees expect to be paid more as their jobs get more demanding. Many employers are planning to run leaner. Something has to give.

4. The first rung of the ladder is disappearing

This was the most consistent warning in the report, and I think the most underestimated. Companies aren't laying off junior people. They've simply stopped hiring them. Stanford payroll research has entry-level employment in AI-exposed jobs running about 19% below trend.

The entry-level jobs that remain have changed, too. PwC found entry-level roles most exposed to AI are now seven times more likely to require traditionally senior skills like leadership and creativity.

That entry-level work is how people learn the business. It's where future managers come from. Cut it, and you save money today while building a seniority cliff a few years out, right as experienced people retire. 78% of HR and talent leaders worry AI could weaken their future leadership pipelines. Few have a plan for it.

5. Judgment is the skill that just got more valuable

When a junior employee and an AI agent can produce the same first draft, the draft stops being the value. The value sits with the person who reads the situation correctly and makes the call.

People are already leaning on AI to fill gaps. Robert Walters found 80% of US workers use AI tools to cover skills or knowledge they don't have, often without much training or guidance from their employer. Knowing when to trust that output is a judgment call.

Most companies have never defined good judgment, let alone measured it. They assumed it came with seniority. Judgment can be trained and tested, and the companies that start now will look a few years from now like they hired better people.

The leadership question should change too. Stop asking whether your people are using AI. Start asking whether they're getting better results because of how they use it.

6. The labor shortage is hiding in plain sight

Low unemployment makes it look like workers are easy to find. Underneath, the workforce is shrinking. The US labor force is about 700,000 people smaller than it was in January, and participation fell from 62.3% to 61.6% in a year. The BLS projects labor force growth of just 0.3% a year over the next decade, down from 0.8% over the last one.

The Dallas Fed now estimates the economy needs roughly zero net new jobs a month to hold unemployment flat. In 2023, that number was about 250,000.

Hiring is slow right now, so nobody feels the squeeze. That changes the moment hiring picks back up. The bigger question for the next five years may be whether AI can fill the jobs we no longer have enough people for.

7. Trust and culture decide whether AI pays off

The companies furthest ahead with AI changed how decisions get made around the technology. Spending more didn't get them there. Money buys capability. It doesn't buy the operating changes that turn capability into results.

Trust runs through this whole report. Employees need to feel safe enough to experiment and make mistakes. Managers need to trust AI output enough to act on it. And as AI agents take on more decisions, someone has to own the outcome when one gets it wrong. Most companies haven't answered that yet.

Be careful reading loyalty into the numbers, too. Quits have hovered near 1.9% for almost two years, the lowest sustained level in a decade. People staying because they're afraid to leave can look a lot like engagement on a dashboard.

Thanks for reading. Want to talk more? Check the form below or email me!