** Wired for Thought **
The stamina required to endure the AI winter and emerge into the present era makes me grateful I didn’t choose academia. Worlds I See offers a reasonable introduction to the modern history of artificial intelligence and neural networks, though its publication date means it omits recent advancements.
One intriguing idea Li introduces, though without deep defense, is her version of Dunbar’s number: how many concepts is the human brain aware of? She suggests 30,000—what could be called ‘Fei-Fei’s number’—as a probable natural limit to human cognition and labeling, aligning with dictionary entries and her goal for ImageNet.
Another compelling perspective is her belief in evolution as the only force powerful enough to create intelligence—flipping creationist arguments on their head. She also thinks the Cambrian explosion was caused by sensory perception, compelling to my neural net!
Li also details her immigrant experience, noting grades are the natural hope for immigrants. At least she had a family that encouraged her in math, unlike many immigrant families!
** 16th book of 2025 **