Would AI have predicted human air flight just before it happened?

In its current state of the art, AI seems not compatible with human creativity, as Felin and Holweg (2024: 28) exemplify with the invention of aviation: given all empirical evidence available at the end of the nineteenth century, AI will not have been able to “predict” the development of the aerospace industry at the beginning of the twentieth century. . .

Félix‐Fernando Muñoz (2024). “The coevolution of technology, markets, and culture: the challenging case of AI.” Review of Evolutionary Political Economy (acccessed on line at https://doi.org/10.1007/s43253-024-00126-0)

I quote at some length from the Felin and Holweg (2024) just referenced:

So, what was the evidence for the plausibility of human powered flight at the time [in the late 1800s and the early 1900s]? The most obvious datapoint at the time was that human powered flight was not a reality. This alone, of course, would not negate the possibility. So, one might want to look at all the data related to human flight attempts to assess its plausibility. Here we would find that humans have tried to build flying machines for centuries, and flight-related trials had in fact radically accelerated during the 19th century. All of these trials of flight could be seen as the data and evidence we should use to update our beliefs about the implausibility of flight. All of the evidence clearly suggested that a belief in human powered flight was delusional. A delusion can readily be defined as having a belief contrary to evidence and reality (Pinker, 2021; Scheffer, 2022): a belief that does not align with accepted facts. In fact, the DSM-4/5—the authoritative manual for mental disorders—defines delusions as “false beliefs due to incorrect inference about external reality” or “fixed beliefs that are not amenable to change in light of conflicting evidence.”

Notice that many people at the time—naïvely, it was thought—pointed to birds as evidence for the belief that humans might also fly. This was a common argument. But the idea that bird flight somehow provided hope and evidence for the plausibility of human flight was seen as delusional by scientists and put to bed by the prominent scientist Joseph LeConte. He argued that flight was “impossible, in spite of the testimony of birds” (1888: 69). Like a good scientist and Bayesian, LeConte appealed to the data to support his claim. He looked at bird species—those that fly and those that do not—and concluded “there is a limit of size and weight of a flying animal.” According to LeConte, weight was the critical determinant of flight. With his data, he clearly pointed out that no bird above the weight of 50 pounds is able to fly, and thus concluded that therefore humans cannot fly. After all, large birds like ostriches and emus are flightless. And even the largest flying birds, he argued—like turkeys and bustards—“rise with difficulty” and “are evidently near the limit” (LeConte, 1888; 69-76). . . .

The emphasis that LeConte placed on the weight of birds to disprove the possibility of human powered flight highlights one of the problems with data and belief updating based on evidence. It is hard to know what data and evidence might be relevant for a given belief or hypothesis. The problem is—as succinctly put by [Karl] Polanyi—that “things are not labeled evidence in nature” (1957: 31). Is the fact that small birds can fly and large birds cannot fly relevant to the question of whether humans can fly? What is the relevant data and evidence in this context? Did flight have something to do with weight, size, or with other features like wings? Did it have something to do with the “flapping” of wings (as Jacob Degen hypothesized)? Or did it have something to do with wing shape, wing size, or wing weight?19 Perhaps feathers are critical to flight. In short, it is hard to know what data might be relevant and useful. . . .

So, what might happen if we weight our beliefs about the plausibility of human flight by focusing on reliable, scientific sources and consensus? In most instances, this is a rational strategy. However, updating our belief on this basis when it comes to heavier-than- air flight during this time period would further reinforce the conclusion that human powered flight was delusional and impossible. . . .

In the case of human flight, the data, evidence, and scientific consensus were firmly against the possibility. No rational Bayesian should have believed in heavier- than-air flight. . . .

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4737265 (my bold)

Why would we belleve today’s AI would have done any better then with that information?