Why Speed Mistakes Us for Talent

Isabella

So, I, I was looking at this, this freelance marketplace data, and it completely blew my mind. Like, we all think we are these rational, objective judges of talent, right? But it turns out we are utterly, hopelessly addicted to speed. In this July 15, 2026 study in Management Science by Eric VanEpps and Einav Hart, they looked at over eleven million transactions on Fiverr. Eleven million! And they found that a candidate who takes just one hour longer to reply to a message is forty-six percent less likely to get the job. Forty-six percent! For one single hour of delay.

Jesper

Forty-six percent? That is, um, that is wild. I mean, forty-six percent less likely because of sixty minutes? That is, yes, that is a massive penalty for just, you know, being away from your phone or actually, uh, doing some real focused work.

Isabella

Exactly! We translate "fast" to mean "competent" or "warm" or, I don't know, "deeply committed to my minor project." It is the ultimate quick-reply bias. We are literally letting the inbox vote on our talent pool.

Jesper

Yes, and it really speaks to what we in Denmark often call mavefornemmelsen. You know, that gut feeling when you are hiring. We have this very informal culture, and we think, oh, we had a great conversation, or they responded so quickly, there is god kemi, good chemistry, right? But that gut feeling is so often just a proxy for who has the fastest notifications turned on. If you want to protect your decision-making, you have to write down your evaluation criteria in advance. If speed is not the absolute key to success for the role, do not let that one-hour delay ruin a great candidate. Yes, it is about being disciplined before you even look at the inbox.

Isabella

Yes! Because otherwise, we are just rewarding who is the most reactive, not who is the most thoughtful. And this gets even crazier when you look at how we measure productivity once people are actually on the job. There was this HBR paper from July 2026 that talked about the performance paradox. They cited an Organization Science experiment with seven hundred and fifty knowledge workers using GPT-4. Now, on normal tasks, the AI-using workers were twenty-five percent faster. No surprise there. But when they gave them a task that was just outside the AI's capability, the AI users were nineteen percent less likely to find the correct solution than the people who were not using AI at all.

Jesper

Nineteen percent less likely? Wow. So, they are, they are rushing ahead, trusting the tool blindly, and they actually perform worse because they do not stop to, uh, verify or think. That is a massive blind spot.

Isabella

Yes! Because our current KPIs are completely blind to that boundary. They see the worker who copy-pastes an AI answer in five minutes as a superstar, and they punish the worker who takes two hours to research, verify, and correct the AI's hallucinations. We are rewarding speed and punishing actual judgment.

Jesper

Yes, and this is where we have to shift from these flat, automated dashboards to what I would call complementarity conversations. In Denmark, we have this strong tradition of tillidsbaseret ledelse, trust-based leadership. We do not need constant, invasive digital surveillance to see if people are typing. Instead, in your weekly one-to-one, just ask one simple question: "When did you last catch an AI error, or reframe a problem that the AI did not see?" That is where the real human value is added. If we only measure output volume, we are just inviting a flood of fast, mediocre nonsense.

Isabella

And that brings us to the really scary part. If we automate all the routine tasks, how does anyone actually learn how to catch those errors? Back in Week 27, we talked about how AI is hollowing out entry-level roles. And the economic data is, uh, it is pretty brutal. In the first quarter of 2026, college graduates faced a five point seven percent unemployment rate, and forty percent were underemployed. And Stanford's Digital Economy Lab showed a sixteen percent decline in employment for young workers in the most AI-exposed fields.

Jesper

Yes, the, the bottom of the talent pyramid is essentially dissolving. Mark Russinovich and Scott Hanselman from Microsoft called this the AI drag for juniors. Seniors get an AI boost, they fly through tasks. But juniors face a drag because they do not have the fundamental judgment to know when the AI is hallucinating or just giving a lazy answer. If we just hire seniors and automate the junior roles, who is going to be the senior in ten years? We have to make junior growth an explicit, deliberate goal, not just hope they learn by osmosis.

Isabella

Right! The old way of learning, where you sit next to someone and watch them work, what we call sidemandsoplæring, it just does not work when everyone is staring at their own chat window. But there are ways to build this judgment on purpose. McKinsey highlighted this medical preceptor model, and there was a fascinating study in the Journal of Surgical Education in October 2025. First-year medical students did this five-day "attempt-then-feedback" loop. They would try to diagnose a patient first, then compare their logic with the AI's diagnostic reasoning, and then dissect the differences with a senior mentor. After just five days, those first-year students outperformed second-year students, who were a full year ahead of them in their training! And the effect lasted when they tested them again two weeks later.

Jesper

That is an incredible acceleration! Yes, we can call this the answer-key model. The junior attempts the task first, independently. Then they run it through the AI, and then the manager talks through the gap. That is how you resurrect the classic Danish mesterlære, the master-apprentice model, in the digital age. You are closing the "Pas på trinet!" gap, the step-by-step learning that Dennis Nørmark talks about. Bank of America is doing this right now with their 2026 cohort of nearly four thousand interns. They are using simulated environments to compress that years-long judgment-building process into weeks.

Isabella

It is so much better than just handing them a polished AI template and saying, "Here, format this." But, Jesper, how do managers find the time to do this? Everyone is already drowning in goals and projects. If we add "intensive mentor sessions" to their plate, they are going to collapse.

Jesper

Yes, you are entirely right. This is about managing cognitive capacity as a finite resource, just like budget or headcount. McKinsey had this great piece in July 2026 about the cognitive cost of change. The big problem is what Brian Heger calls goal creep. We just keep adding new initiatives, new software, new goals, without ever taking anything off the plate. It is a quiet killer of mental bandwidth.

Isabella

Totally. It is like, "Hey, here is a new AI tool and a new tracking system, but also, keep doing everything you were doing before."

Jesper

Yes! And in Denmark, we have a great word for the stuff we should be getting rid of anyway, pseudoarbejde, pseudo-work. Tasks that have no real value but look busy. So, my challenge for leaders is to institute a mandatory "stop meeting." Before you launch any new project or initiative, the team must sit down and agree, in writing, on exactly what work stops to make room. If you can't name what is going to stop, you don't get to start the new project. Yes, it is that simple. Protect that bandwidth so your people actually have the cognitive space to think, to verify, and to mentor the next generation. That is how we rebuild the ladder.

Isabella

I love that. No stop list, no new project. Well, that is a perfect place to leave it for today. Thanks, Jesper, and we will talk next week.

Jesper

Yes, thanks, Isabella. Take care of each other out there. Bye.