“To make discoveries, we need to reward, not punish, those who risk breaking away from the crowd,” says Avi Loeb, head of the Galileo project and founding director of the Black Initiative, in a recent article. Hole” of Harvard University. university.

In his analysis, published on Medium, Loeb draws on a statement about human regression made by psychologist Daniel Kahneman when he won the Nobel Prize in 2002. because we reward others and get rewarded for punishing them,” said the late psychologist.

Thus, Loeb notes, it is human nature for some scientists to make it their mission in life to “step on every flower that rises above the level of the grass in their field of study.” Similarly, those who perform below expectations are those who demand “performance parity.”

Regarding this “regression to the mean” referred to by the psychologist, Loeb points out that the perception of the “mean” when analyzing people is wrong, and the only way to correct this is to examine deviations from the mean. “In order to make discoveries, we need to reward, not punish, those who risk standing out from the crowd,” the researcher emphasizes.

This regression can also be traced through the analysis of some AI tools that rely on human input to generate responses, such as ChatGPT – 3.5 and GPT4, which have regressed over time.

A study comparing the ability to solve math problems, answer sensitive questions, generate code, and perform visual reasoning in March and June 2023. For example, ChatGPT4’s response accuracy was found to have dropped from 97.6% in March to 22.1% in June.

The way people treat academic performance, including in the university environment, is the cause of this regression. In this sense, he recalls a joke that he says he often uses at his workplace.

“Harvard’s top students are exceptional because they get 10s in all their courses, while Harvard’s worst students get 10s in all their courses,” notes Avi Leob.

Regarding such practices, the researcher insists that they result in an average placement of all people without deviations, but in fact represent a lack of value because the grading system was created to differentiate students according to their performance.

What would it be like, the researcher asks, if such practices were transferred from academia to a field based on competitiveness, such as sports.

“Imagine the Olympics, where the officials slow down the fastest 100m runners so they finish with the slowest runner. This would be a great gesture for the slowest runner, but it hides the inherent diversity of athletic ability. It undermines the idea that the Olympic Games are a celebration of attitudes that go against psychological habits,” says Loeb.

If he were to compare Kahneman’s quote with his example from the world of sports, the researcher says it would be like this: “on a 100-meter track, a group of runners shows a progression that is far from average; as part of the Olympic spirit, we celebrate those who run the fastest, instead of rewarding the slow and punishing the first.”

Finally, he notes that performance was the main criterion when he began his research career, just like sports, and the system must understand that breakthroughs in science are deviations from the average.

“Supporting these findings, academia and academic culture have resisted the natural psychological tendencies of regression to the mean,” he concludes.

• Avi Loeb is the head of the Galileo project, founding director of the Black Hole Initiative at Harvard University, director of the Institute for Theory and Computation at the Harvard-Smithsonian Center for Astrophysics, and former dean of the Harvard School of Astronomy (2011–2020). ). He is a former member of the President’s Council of Advisors on Science and Technology and a former chairman of the Council on Physics and Astronomy of the National Academies.
• He is the author of the bestselling book Extraterrestrial: The First Sign of Intelligent Life Beyond Earth and co-author of the textbook Life in the Cosmos, both published in 2021. His latest book, Interstellar, was published in August 2023.