Happy Friday,
What advice would you give a 21-year-old responsible for hiring three people? I am asking sincerely, because that 21-year-old is me, and I have been turning the question over for two weeks. It sits at the center of everything this newsletter believes: as machines take on more of what we do, what does a human being still bring to a team? This week the question stopped being theoretical, because I have three seats to fill and real expectations attached to them.
The past two weeks have been chaotic and travel-heavy. The rest of the founding team and I announced that Moodr Health raised a $3 million seed round in health tech, and we have been in California and Colorado meeting investors and potential partners. Alongside the excitement came a quieter reality: investment carries expectations about how we use the money, and I have been tasked with bringing three more people onto our team.
That responsibility made me step back. What is the value of a human on a team? What matters beyond the functional work someone performs?
What if an agent could do the job?
Through my studies in artificial intelligence, I caught myself asking a question I did not love: could I replace a role with an AI agent? What if I put the $100,000 I might spend on an annual salary toward developing, deploying, and using an agent instead? I started a pros-and-cons list, and then I had to check myself, because the comparison treated a person as a line item of output and overlooked so many inherently human components of being on a team.
What do people bring that agents do not?
I hire a human because I want their human experience and connection. Being on a team involves much more than productivity. If we are going to spend 40 hours a week working together in this capitalistic environment, I want to spend that time with SOMEONE. Culture absolutely matters.
A person brings a life into the room: judgment formed by things that never happened at work, loyalty built in hard weeks, the ability to sense that a teammate is off before any metric shows it. An agent can produce an answer, but a team runs on more than answers. I want people beside me who know what a hard week feels like and can read the silence in a room, and I do not believe that kind of understanding can be delegated to software. Human prowess, the thing this newsletter exists to defend, shows up precisely in those moments, and the future of teams will be shaped by whether we still know how to value them. I am betting on people, and I want the human experience of work to stay human-centered even as our tools get astonishing.
I hire a human because I want their human experience and connection.
What did the research miss about human-AI teams?
The research world just gave this question a fascinating new chapter. In 2024, a widely covered meta-analysis in Nature Human Behaviour reviewed 74 studies of human-AI collaboration and landed on a sobering result: on average, the human-AI teams performed worse than the better of the two working alone. That finding traveled widely, and it fueled real doubts about whether human-AI collaboration was worth pursuing at all.
Then this month, researchers at the Max Planck Institute for Human Development published a reanalysis of all 74 studies in PNAS, and the picture changed. Their argument: the original studies systematically underestimated human learning, and most never examined it at all. Only 10 of the 74 gave participants any feedback on whether their answers were right. When people could actually learn, with both outcome feedback and explanations from the AI, the human-AI teams tended to outperform the better partner alone. Explanations without feedback, on the other hand, were associated with worse performance. One honest caveat: none of the 74 studies varied feedback experimentally, so these are associations across studies, not causal proof. The authors say that plainly and call for experiments that test learning directly.
FROM THE LITERATURE. Human learning as the missing variable in human-AI teams. A reanalysis of 74 studies behind the famous finding that human-AI teams underperform the better partner alone found that the result flips when people can learn: outcome feedback paired with AI explanations was associated with teams outperforming the better partner alone, while explanations without feedback were associated with distinctly negative synergy. Only 10 of the 74 studies offered any feedback at all, and none varied it experimentally, so the authors call for human learning to be treated as central and for causal experiments to follow.Berger, J., et al., "Human learning is an understudied but promising lever for boosting human-AI synergy," Proceedings of the National Academy of Sciences, 2026
Two cautions from the authors deserve to travel with the finding. First, a chatbot's answer is not feedback. A chatbot can be wrong, and its answer does not tell you what was actually correct; real feedback is knowing the outcome. Second, offloading. Delegating our mental work to AI can quietly erode the learning it was supposed to support, and over time it can cost us skills we never noticed losing.
For me the practical takeaway is personal: as we use AI, we have to keep exercising critical thinking and developing our own thoughts, opinions, creativity, and human taste. If the research is right, the advantage goes to teams whose people keep learning inside the partnership, checking the machine against reality instead of handing the thinking over. That standard applies to me as much as to anyone I hire.
Where does this leave me?
Honestly, still in the middle of it. I have three roles to fill, a team I believe in, and a standard I am still learning to articulate. I am hiring three people whose judgment, energy, and presence will shape what Moodr Health feels like to work at, and that feels weightier than any job description I could write.
So I will end where I started, because many of you have built teams, hired well, hired wrong, and learned the difference the hard way. What advice would you give someone who is 21 and now has to hire? What would you tell me?
Bet on people,
See you in two weeks,
Jaron
The minds we pulled from this week:
Berger, J., et al. (2026). Human learning is an understudied but promising lever for boosting human-AI synergy. Proceedings of the National Academy of Sciences. https://www.pnas.org/doi/10.1073/pnas.2536100123
Max Planck Society. (2026, October 15). Human learning is an underestimated factor in the success of human-AI teams. https://www.mpg.de/27037803/human-learning-is-an-underestimated-factor-in-the-success-of-human-ai-teams
Vaccaro, M., Almaatouq, A., & Malone, T. W. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour. https://doi.org/10.1038/s41562-024-02024-1