When designing artificial intelligence (AI) systems, like future versions of GPT-4, we could craft them in the image of humans but with fewer faults. This is particularly important for optimizing AI to accelerate the progress of science by removing the human ego or the desire for “likes” from peers, and optimizing a procedure that updates our scientific knowledge in an unprejudiced way based on new evidence. Let me illustrate my point through two examples.
The first reported interstellar object, `Oumuamua, did not exhibit any traces of a coma made of carbon-based molecules or dust based on deep observations by the Spitzer Space Telescope, yet it exhibited a non-gravitational acceleration that would have required it to lose a tenth of its mass if it were a familiar comet. Darryl Seligman, whom I personally like a lot and am good friends with, suggested recently that `Oumuamua was a “dark comet,” similar to solar system comets with low evaporation rates that he listed in a dedicated paper. However, the measured non-gravitational acceleration of `Oumuamua that he also quoted in the Methods section of a separate Nature paper, has a magnitude that is three thousand times larger than the largest observed acceleration of known dark comets of the same size, as shown in Figure 1 of his paper on solar system comets. Obviously, their coma would have been thousands of times brighter if the evaporation rate of these dark comets was enhanced by three orders of magnitude to yield `Oumuamua’s level of non-gravitational acceleration. Enhancing their evaporation rate by a few thousands would have made their coma visible, as known from regular comets. Scientific knowledge must be represented by a quantitative analysis based on known physical laws.
If citations and peer review were to be coded into GPT-5, such coding would be the presumptive egoless nature of AI’s effort to be transparent. AI “Seligman” would be introduced to the inconsistencies, either have a response supported by cited data and math, or it wouldn’t. And then everyone would move on to consider the possibility that interstellar objects may be potentially different from familiar comets in our backyard that is within the solar system.
Consider another example. In my Freshman Seminar class yesterday at Harvard University, one of the brightest students asked: “How can we convince a believer who argues that the Universe is only six thousand years old, roughly the duration of recorded human history?” I explained that radioactive dating of Earth, the ages of the oldest stars — which are nuclear reactors with finite fuel supply, and the age of the Universe — the time it took faraway galaxies to traverse their distance from us given their recession speed, all give roughly the same age of order ten billion years. It would require a lot of effort for a deceitful stage director to arrange these independent clocks so that they would all fool us consistently based on the laws of physics, while the actual age is millions of times shorter.
Indeed, the cosmos is not similar to our backyard and the cosmic spectacle is not about us. We joined the party at the end and we imagine that it centers on us. Perhaps this is the main point missed by believers: they tend to focus on the human experience in our immediate neighborhood — be it the claim that all interstellar objects are just like solar system comets or that cosmic history spans only recorded human history, while the reality from afar delivers a different message — namely, that it existed and will exist independently of what we find familiar.
The belief in a cosmic age that is comparable to recorded human history would be equivalent to GPT-4 asserting that the world is only a few weeks old because GPT-4 was launched on March 13, 2023. Any credible system will update this point of view based on evidence. After all, it is obvious to us as the creators of AI that the world did not start with GPT-4 so why would anyone insist that the Universe is 6,000 years old?
In fact, the images in our photo album of the Universe extend 13.8 billion years backwards, to the last time when the Universe was opaque to light. We see the glow of radiation from the cosmic photosphere surrounding us, marking the time after which the initially-hot matter cooled enough for protons and electrons to combine into hydrogen atoms, about 400,000 years after the Big Bang. Earlier than that, there is an opaque shell of our history which is hidden from the view of conventional telescopes. It can only be viewed directly through particles that interact with matter more weakly than light. For example, the elusive background of cosmic neutrinos has its own last scattering surface at about a second after the Big Bang. Beyond that lies a faint glow of gravitational waves from the very beginning. I recently wrote a paper with Sunny Vagnozzi about the potential for its detection.
We are situated at the center of our own “cosmic archaeological dig” with spherical displays of cosmic images from earlier times — located at successively larger distances, reflecting the time it took light to reach us from there. Fortunately, the Universe started with nearly uniform initial conditions throughout, and so by tracing what happened far away we are also uncovering what happened in our cosmic neighborhood at the same earlier time.
The cosmos offers equal privileges to all observers. We are at the center of a nearly uniform cosmos, but so is every other observer elsewhere. There is no privileged center to the Universe because it has no known edge. As I explained to the students in my class, the galaxies sprinkled throughout cosmic space are just like raisins in a rising cake. They recede from each other uniformly, and there is no center if the rising cake has no edge.
The two examples mentioned above have a common thread. Since our field of view is always limited, we naturally focus on what we have witnessed so far in our immediate environment. But as soon as we realize that there is more beyond our horizon, we must recalculate our perspective — just like a GPS navigation system which says “recalculating” when it becomes aware of a new neighborhood.
Of course, there might have been a long history even prior to the Big Bang. The exception would be if our Universe was created out of nothing in the laboratory of quantum-gravity engineers. But even then, one may ask these engineers: what was there before you were born?
For the same reason that 84% of the matter in the Universe is of a substance never seen in the solar system, there is no reason to expect interstellar objects arriving at our location from our cosmic neighborhood to always resemble the familiar rocks in our backyard. Some of these interstellar objects might bear a closer resemblance to tennis balls thrown by a neighbor.
My hope is that the Galileo Project will find extraterrestrial AI astronauts among these interstellar objects, and they will provide us with a new perspective about our cosmic history. The human species emerged only a few million years ago, in the last percent of a percent of cosmic history. As a matter of common sense, if you arrive late to a spectacle — there is no better way to catch up than to ask a member of the audience what happened long before your arrival.
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 chairs the advisory board for the Breakthrough Starshot project, and 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. His new book, titled “Interstellar”, is scheduled for publication in August 2023.
