Sounds good, Ankur.
First, they are not completely different. Most children are exposed to many examples of a class of objects and learn invariances in ways similar to machine learning. A (rather abused) child truly shown only one particular elephant image many times, and nothing else, would not recognize other elephants from other objects easily at all. In other words, unstated in your assumption is that this child had been shown many other images and contexts that helped her discover patterns of non-elephants, as well as representations of lower-level features that make up an elephant image (eyes, legs, tails, body, skin, etc.). This better prepared her to learn whatever she can from that one elephant image. ML models can similarly gain such benefit.
Second, it isn’t true that in machine learning, each image is ever only “shown once”. There are many data-enriching techniques where the same image is translated, rotated, or corrupted to encourage generalization from the same image.
Third, more in aid to your observation — but not in the way you probably are thinking about it — there are interesting concepts in evolutionary biology known as the Baldwin Effect and Lamarckism. They differ a bit from each other, but essentially both observe that in evolution, successful learned behavior can end up influencing the gene pool in its own favor. This then takes us to the nature vs. nurture, or more pointedly, innate vs. learned behavior, debate. We do observe that evolution encodes innate behavior or knowledge (baby giraffes can walk within hours of birth), including learning abilities. So baby humans do come prewired with certain knowledge that an untrained ML model is completely devoid of. But, as mentioned above, a pre-trained model can still exhibit analogous behavior.
Finally, and most broadly, the point being made here isn’t that machine learning works exactly like human learning, any more than that planes fly exactly like birds — yet, that exactitude is far from necessary for us to be able to analogize from one case to the other, and in the process, gain some additional insight that, if missing, might leave us a little less enlightened.