The Human Edge: Why AI Should Make People More Valuable, with Steve Cadigan

Steve Cadigan was LinkedIn's first Chief Human Resources Officer, arriving when the company had roughly 400 employees and helping it scale past 4,000 across 17 countries during the most transformative stretch of its history. Before that he had worked across about six industries, from fashion at Esprit in the mid eighties to insurance to high tech, including four years at Cisco where he worked on close to 50 acquisitions around the world. Today he advises organizations including Google, Salesforce and the BBC through Cadigan Talent Ventures, and he wrote Workquake to argue that the world of work has changed faster than our assumptions about it.

The premise of this episode is a direct challenge to how most companies are deploying AI. The default move is to use it the way every previous technology was used: do more, go faster, save money. Steve's argument is that using the most capable technology we have ever had in order to subtract people, rather than add capability, is a strategic error, and one that compounds.

Across the conversation with host Darcy Mehta, he makes the case for repurposing rather than replacing, explains how LinkedIn beat far richer competitors for talent without outbidding them, and lands on a reframing many people leaders will find uncomfortable: stop trying to keep your best people, and start trying to make them better.

Why using AI to subtract is a race to the bottom

Steve is blunt about why organizations behave the way they do. They are using AI like they used every prior technology, because no new playbook exists yet and the old one is sitting right there. Leaders are tired, anxious and worried their competitors are moving faster, so they reach for the shortcut.

His challenge is commercial rather than moral. If AI can absorb a large share of the work your people currently do, the interesting question is what those people could do next: discover new markets, apply different skills, serve customers in ways you could not previously afford. Eliminating them instead saves money now without answering how you intend to make more later.

He also points at an internal contradiction most HR leaders will recognize. The first team cut when times are tough is training and development, at precisely the moment organizations say they need new skills faster than ever. And leading with fear carries a predictable cost: frighten people with AI and everyone starts playing career defense instead of offense, which slows the company down exactly when it needs to move.

How LinkedIn won talent it could not outspend

The irony Steve enjoys most is that he ran talent at LinkedIn: a company whose product is a recruiting tool, whose biggest constraint was recruiting, and whose competitors could outpay it many times over. Google, he notes, was making more money in a day than LinkedIn made in a year. Outbidding was never on the table.

So the team asked a question he had never asked in more than 20 years in HR before that point.

Not why should they, but why would they, given every other option in front of them. The answer that surfaced was the problem itself: helping people find their dream jobs. The second answer was structural. If LinkedIn was going to call itself the network for professionals, it had no business being anything less than the best place a professional had ever worked. That became an explicit commitment from the executive team, and it changed how interviews opened: where do you want this job to take you, and how can I help you get there.

His warning for anyone copying the move is that the answer has to be genuinely unique. Nice, good leadership, interesting work and competitive pay are what everyone claims, and in a transparent market, where people can see every other choice, sameness does not persuade anyone.

The skills shortage is a development failure we created

Steve rejects the framing that the skills shortage is an AI story. Organizations need people with new skills faster than ever, and most have gutted the very infrastructure that would produce them, treating learning and development as a perk rather than a core business process.

The clearest evidence is the disappearance of graduate and internship programs. He visits college campuses around the world and describes the look on students' faces. His objection is strategic rather than sentimental: this is the generation that went through COVID in school, expects things to be broken, and arrives with a hacker mentality that assumes it can fix them. Locking them out while complaining about a skills gap is a contradiction.

Behind it sits a question he thinks leaders should answer honestly. Is your company being built to last a hundred years, or a hundred months? Many of the leaders he talks to are privately planning on a five to ten year horizon, which makes disinvesting in development perfectly rational. His counter is simple: if you are not building the people who will run the company in five years, who is doing it for you?

Why calling AI a tool stalls adoption

Asked why 88% of employees use AI at work while only a small fraction use it in genuinely transformative ways, Steve points at the language senior executives use. Calling it a tool understates what it is, and if leadership frames it that small, adoption stays that small.

The deeper blocker is self-interest. If people suspect the technology could render them less valuable, they have no reason to accelerate it. That is a rational response to a badly framed rollout, not resistance for its own sake.

His prescription is that every company should write an AI philosophy, a governing principle sitting alongside its stated values, then check that the two actually agree. He pairs it with a memorable test: give every AI project a food label, the way a can of soup carries one. What is in it, what does it add, what does it subtract, does it put people in or take people out, and which privacy or ethical additives has nobody looked at yet.

He is direct about the specific risk in HR. AI tools now make it trivially easy to surface things about candidates that should not be surfaced, and the guardrails do not exist yet. In his view the companies building these products carry more of that responsibility than they have so far accepted, because no government body is close to matching their sophistication.

Career security beats job security

The heart of the episode is Steve's reframing of retention. Tenure has declined steadily for twenty years, and when he asks audiences whether they expect that to reverse, nobody says yes. If short tenure is permanent, then the strategy most organizations still run is defense.

His model comes from college sports, where the rules changed and every major program now assumes its best player stays a single year. This past season Michigan won a national championship in basketball with nobody from the previous year's starting five, and Indiana won in football with 60% of the roster new. Those programs did not fight the churn. They got dramatically better at onboarding quickly, building cohesion and chemistry fast, and simplifying their systems so players could contribute without three years of learning.

Translated into HR, the move is to say the quiet part out loud. I know you are probably not going to stay forever, so I am going to make this the best learning experience you have ever had, and when you go, I will celebrate that you got a better job, and you will send your friends here. He offers the counterexample too: Clemson, where almost nobody leaves, which hints that making people better is also the thing most likely to make them stay. Either way, retention strategy stops being about locking people in.

The one non-negotiable for the next 90 days

Pushed to name a single action every CHRO should take, Steve does not reach for a policy or a platform.

What people are asking for, he argues, is not a promise of job security. It is: make me better for an uncertain future. And his definition of how to deliver that is deliberately unglamorous. Not a catalogue of classes or tuition reimbursement, but learning built into the job itself, through new challenges and new opportunities that let people grow while the business adapts around them.

What the benchmark data says about rewards

Steve's point that rewards is not the lever is borne out by our own data. In CultureMonkey's benchmarks, Rewards is the lowest-scoring named engagement driver globally, and no industry averages above 4.10 on it. Organizations are competing hardest on the ground where they are collectively weakest, which is exactly the trap he is describing. Our benchmark set measures engagement drivers rather than tenure or intent to stay, so it speaks to why rewards is such difficult ground rather than to what replaces it. On that narrower question the data is unambiguous, and it points the same way he does.

What continuous listening is really for

On listening, Steve's warning is that the act of asking creates an obligation. Listening is easy and acting on it is hard, and managers who hear "employee survey" and think "more work for me" have usually been trained to think that by surveys that went nowhere.

He also inverts the instinct of perfectionist leaders. The point is not only to find what is broken, but to find what is working so you do not accidentally stop doing it, and to replay that back to the people generous enough to tell you.

His most useful reframe is about ownership. When a leader sees something broken, the instinct is that they have to fix it. He argues the job is to help get it fixed, which turns the loudest voices in the survey into the people helping solve the problem.

He closes with a story from LinkedIn. A vocal group of San Francisco employees wanted a free shuttle because Google, Facebook and Apple all ran one. Rather than refusing, he went and got the data: two buses each way would cost a quarter of a million dollars, and the San Francisco contingent turned out to be a small minority, with more employees living further away in other parts of the Bay Area. He brought that back and asked whether it was a fair use of the money. They said no, and thanked him for looking into it. Months later, when the population shifted, they got the bus. The lesson is that a survey is a trust-building instrument, and that explaining a no can earn more than granting a yes.

That is where pulse surveys and continuous listening earn their place in this argument. If talent really is loyal to learning, the thing worth measuring is whether people believe they are getting better, and that signal shows up long before it appears in your attrition data.