The End of Learning

Learning is the fundamental process of life. For example, a dog learns to sit when trained. At the same time, a man learns how to train dogs. Dogs learn to be good pets over the course of a lifetime, but that’s on the tail end of a much longer process by which species of dogs learned the traits that would make them more fit for domestication.

Neoteny is the trait whereby some animals, like beavers, are born very immature. The evolutionary advantage that comes with this is newborn animals are given the opportunity to be taught by their mothers. This is essential for beavers, since they’re nature’s craftsmen. For mankind, our neoteny in a sense lasts our whole lives. Because our way of life is so well insulated from primitive nature, we’re able to keep learning from each other all our lives.

Humans and animals learn over the course of their lives, but before that, they learned to learn through evolution. The incredible strategy that emerged from evolution is that a species doesn’t need all the specific strengths that will make it more fit than the rest in order to win. It’s enough for the human species that we’re the most capable at adapting and learning new skills.

After we won the game of evolution, our species-level learning moved into culture and technology that outlasts an individual lifetime. Nonetheless, our species-level learning serves to accelerate our capacity to learn at the individual level, over the course of a lifetime. In this way, learning is a circular phenomenon. Since humanity emerged from natural selection, the telos of our collective learning has been a greater capacity for individual learning.

Despite being insulated from nature, our evolutionary project continues. Each generation of humans is able to learn more efficiently, because we collectively advance our ability to learn. At the same time that technological evolution gives us greater leverage in learning, it also crystallizes and immortalizes the intelligence we have so far.

But technological evolution is approaching a conclusion similar to when natural selection landed on the human specimen. AGI is a unique technology because it internalizes all of our collective intelligence. It’s also very different from humanity. Whereas humanity was selected for our capacity to learn, AGI will not need to learn after training. AGI will be selected through the quasi-evolutionary process of training to have all of the specific capabilities we have, but none of the adaptability.

At this point it seems clear that once AGI arrives it will not continue to learn. As for us, our individual learning will be better than ever, but our collective learning will come to an end. AGI is, and perhaps always was, the end of learning.