AI’s opportunity paradox

More help, fewer opportunities

Will AI empower or displace knowledge workers? I feel like every week turns up new hot takes on AI’s labor market impacts. But taken together signals are anything but clear.

Anthropic’s Dario Amodei has warned that half of entry-level work could get wiped out by the AI revolution. Amazon’s Andy Jassy says AI will eliminate some jobs and create new ones, but net out to a leaner workforce. Others, like LinkedIn founder Reid Hoffman and Nvidia CEO Jensen Huang, remain optimistic that AI won’t destroy jobs; it will transform them.

These conflicting forecasts make one thing clear: if you want to future proof your career, employers want you to run with AI, not from it. As Jassy told employees last year, “use and experiment with AI whenever you can.” And just last week, a leaked internal memo at Meta revealed the company’s new incentive bonuses rewarding employees who use AI.

While rational, I worry that these directives have hidden tradeoffs for employees themselves. In a rush to build AI skills to keep their jobs, workers could risk slowly shrinking the very thing they need to get jobs: a broad, diverse network.

Getting a job has always been a social phenomenon. An estimated half of jobs and internships come through networks—and not through our closest friends and loved ones, as you might expect, but instead through acquaintances, or “weak ties.” The further you network away from your friends, the closer you get to your next job.

That’s what Stanford sociologist Mark Granovetter discovered back in the 1970s when he set out to study job-getting patterns among a sample of 282 men. Contrary to popular wisdom about nepotism and networks, when asked how they’d secured their jobs, it wasn’t their uncles or close friends who had hired them. Rather, people they barely knew told them about job opportunities or took a bet on them.

The “strength of weak ties” was confirmed in a more robust study of over 20 million LinkedIn users in 2022. Researchers found users were most likely to find jobs through their “moderately weak ties”—those connections with whom they shared around ten mutual connections. The effect was especially pronounced among job seekers working in the digital economy.

As workers double down on AI, these humble yet vital connections are facing a test: in many ways, interactions with Generative AI can easily outcompete interactions with weak ties. In Microsoft’s Work Trend Index, when asked why they turned to AI instead of a colleague, 42% of workers lauded its 24/7 availability, and 30% said its machine speed and quality. Another survey found that approximately half of Gen Z users actually prefer turning to ChatGPT, rather than their managers, for career advice. More active AI users are even more likely to favor it: A recent Upwork survey found that more than two-thirds of high-performing AI users say they trust AI more than their coworkers, and 64% say they have a better relationship with AI than with human colleagues.

These preferences are rational. Why turn to a colleague who could take hours to respond when AI can give you instant guidance? Why reach out to an executive or CEO who might not give you the time of day when you can get a halfway decent answer at the press of a button? Why burden your boss when you can look smart without even asking for help? Even the best human mentor can’t offer constant, on-demand help with a chaser of infinite (sycophantic) patience. And why do any of these things if employers are more inclined to incentivize employees to partner with AI than with each other?

But AI’s unerring support could gradually atrophy our ability to articulate our thoughts under pressure and navigate complex or tense conversations. It’s 24/7 availability could make convenience of talking to a bot trump connection built through human conversation. But those very human conversations, as frustrating or time-consuming as they may be, add up to the professional networks that outlast jobs. If AI disrupts conversations, it eventually disrupts networks.

What makes this so tricky is that AI’s competitive advantage over our weak ties exhibits all the hallmarks of the Innovator’s Dilemma—when low-margin pursuits are overlooked despite their long-term benefits. The benefits of weak ties are rarely felt in the near term. In our day-to-day lives, we don’t need our weak ties to get by. You don’t miss weak ties until they are missing—when you go out to find a new opportunity or job and find yourself applying in an abyss.

Luckily, disruption isn’t destiny. As individuals, we can start to ask ourselves: How often do I use AI instead of emailing a colleague or making a phone call? When’s the last time I reached out to a new contact to talk about a project I’m working on or a topic I’m interested in?

Eventually I suspect that organizations that manage to pair more individual productivity using AI with stronger social and collaboration habits will have an edge. Until then, workers will have to strike this balance on their own–delivering on their bosses’ productivity expectations while still investing in their own futures–one human conversation at a time.

AI is rewriting the rules of the job market. Yes, AI skills are rapidly becoming table stakes. But the opportunity equation is still about skills + networks. In fact, relationships are becoming increasingly important in securing a job, as employers, overwhelmed by AI-generated applications, turn to referrals—weak ties—to fill vacancies.

That means workers need to consciously choose to stay connected, despite Generative AI’s gravitational pull in the other direction. If we don’t, our skills built and time saved could add up to opportunities lost.