Babies that AI Cannot Create (as Yet)
When I started my career in astrophysics at Princeton in 1987, my mentor John Bahcall asked me for a list of the computer codes that I developed during my PhD. I humbly informed John that I only use computers rarely, when I need to solve a complex set of ordinary differential equations that have no analytic solution. He followed up by asking: “How do you plan to have a successful career in science without developing the skills for code writing?”
At that time, John was not aware of my most important skill. My mind is constantly bubbling with new ideas. After arriving at Harvard in 1993, I attracted a steady cohort of 8 PhD students every year for nearly three decades and gave each of them an original idea for a research topic. During that period, I had numerous instances in which students, postdocs or faculty would arrive at my office to explain what they are working on and I asked them if they thought about a specific new way to approach their research problem. Very often, the response was: “I had not thought about that. This is a very interesting idea.” And in many cases, they ended up working for years on the idea that I proposed shortly after meeting them.
To me, this recurring phenomenon is strange because these ideas arise instantly in my mind without much effort. The question as to why these ideas were not abundantly familiar to others remains a mystery. As far as I am concerned, these ideas are obvious and a matter of common sense, but surprisingly these babies were never born before. In many instances, the ideas seem trivial in retrospect, like establishing a LIGO detector on the Moon where seismic noise is minimal. As a result, the scientists who benefit from working on them end up underappreciating the first few seconds of their birth in view of the years of follow-up work needed to bring these ideas to maturity. This is completely understandable, as the work involved in raising children is far more demanding and less enjoyable than the process of conceiving babies. Nevertheless, we all recognize that adults would not exist if babies were not conceived. The art of defining original research directions is underappreciated by those who tend to work on established research themes.
I was reminded of this recurring pattern in my scientific career just yesterday, when one of my former brilliant postdocs, Dan D’Orazio — who currently works at the Space Telescope Science Institute, came to visit my office. He told me about the planned use of a coronagraph in NASA’s Roman Space Telescope for imaging exoplanets by blocking off the light from their host star. As soon as he mentioned that, I suggested that the coronagraph could also be used to block off the light from quasars and measure the angular separation of their broad line region. Inferring the characteristic velocity in this region from the spectral width of the lines, can be used to measure the mass of the quasar black hole. In case the black hole mass is measured by other means, this method can be used to measure the distance to the quasar, offering a new yardstick in cosmology.
Shortly afterwards, Professor Andrew Vanderburg showed up in my office with a Harvard undergraduate student who wishes to constrain the age of the Universe. We discussed various traditional methods, like isotope dating or stellar ages, but they all suffer from uncertainties that compromise their usefulness in setting novel cosmological constraints. I therefore suggested using clocks made of pairs of white dwarfs that coalesce through the emission of gravitational radiation. Only pairs which are close enough to coalesce within the age of the Universe could be depleted. The planned LISA gravitational wave detector will be able to observe the gravitational wave emission by the population of white dwarf binaries in the Milky Way galaxy. If it detects a step in the distribution of depleted binary separations, that feature can be used as a clock to measure the age of the Universe.
38 years after my conversation with John Bahcall, yesterday’s discussions confirmed that I still possess the skill that enabled my career in astrophysics. It is unclear why some of these baby ideas were not born earlier, but this surprising fact allows me to still make a meaningful contribution to science.
Get Avi Loeb’s stories in your inbox
Join Medium for free to get updates from this writer.
By the end of 2025, code writing is being taken over by artificial intelligence (AI). If an avatar of John would have hired an avatar of me for a five-year postdoctoral fellowship now, he would not have asked the same question. Instead, he would have advised: “To distinguish yourself from the crowd of science practitioners in the coming decades, you will need to come up with original ideas.” Perhaps I was born 38 years too early, but I certainly appreciate the fact that it would take much more time for AI agents to imitate creativity than for them to write computer codes.
***
When an idea is born, it appears as a low-resolution image of an object. Over time, as more research is dedicated to the idea, the details of its terrain are recognized. Sometimes, those who lived through the first phase do not survive to witness the second phase.
On October 9, 2025, the MAVEN (Mars Atmosphere and Volatile EvolutioN) spacecraft took a low-resolution image of the new interstellar object 3I/ATLAS from a distance of 56 million kilometers. This image, captured by its Imaging Ultraviolet Spectrograph, revealed the presence of hydrogen atoms in the vicinity of 3I/ATLAS.
On December 6, 2025, NASA reported here that MAVEN lost communication with ground stations on Earth (see also the discussion with Rob Finnerty on Newsmax here). Telemetry from MAVEN showed that all of its systems were working normally the last time it entered its orbit behind Mars but subsequently NASA’s Deep Space Network did not observe a signal. The MAVEN spacecraft entered Mars’ orbit in September 2014, aiming to explore the upper atmosphere of Mars and its interactions with the Sun. The spacecraft also served as a communications relay station for rovers on the Martian surface, such as Curiosity and Perseverance. Gladly, this task is also handled by the Mars Reconnaissance Orbiter, which continues to operate as usual.
ABOUT THE AUTHOR
Avi Loeb is the head of the Galileo Project, founding director of Harvard University’s — Black Hole Initiative, director of the Institute for Theory and Computation at the Harvard-Smithsonian Center for Astrophysics, and the former chair of the astronomy department at Harvard University (2011–2020). He is a former member of the President’s Council of Advisors on Science and Technology and a former chair of the Board on Physics and Astronomy of the National Academies. He is the bestselling author of “Extraterrestrial: The First Sign of Intelligent Life Beyond Earth” and a co-author of the textbook “Life in the Cosmos”, both published in 2021. The paperback edition of his new book, titled “Interstellar”, was published in August 2024.
