In nature, bigger brains don’t necessary mean smarter creatures. Does this apply to artificial brains, too?
In What Technology Wants, Kevin Kelly argues that the relentless increase in artificial brains’ size doesn’t necessarily mean machines will become smarter or sentient.
To make his case, Kelly looks for parallels in the organic world, a recurring theme throughout the book. Consider the brain of a sperm whale. With an average of 8 kilograms, a sperm whale’s brain is at least 100,000 times bigger than an ant’s. Individual ants are far from clever, but are they 100,000 times less smart than a whale?1
Human brains average 1.4 kilograms. No offense to the whales, but we can all agree that they are not 4 times smarter than us.
Bigger doesn’t mean smarter, even when looking at brain-body mass ratio instead of mere mass. Small ants have a brain-body mass ratio of 1/7, humans 1/40 but humans are unquestionably more intelligent.
Size, clearly, is not the only factor determining intelligence2.
What separates humans from other mammals is a combination of hardware, the brain, and software, the mind that emerges from it. The relatively bigger brain and its structure are instrumental but not the only factor in our intelligence. Our evolved creativity and ability for syntactic language play a crucial role, and there are undoubtedly other components that we haven’t discovered yet.
“The correlation between the absolute scale of the brain and smartness is not significant,” Kelly writes, and the rule might extend from organic brains to digital ones.
Modern LLMs are trained with hundreds of billions, sometimes trillions, of parameters. They can perform remarkable feats and have achieved uncanny levels of human parroting. It’s reasonable to expect they’ll get even more refined as the datasets grow. But that is not grounds for expecting them to give off a spark of Artificial General Intelligence.
LLMs can be more prolific and accurate than a person, but they are never creative in the way a human is. ChatGPT won’t stop replying to your queries because it would rather write its own poetry instead. Creativity and agency are at the core of what makes an intelligence general.
Whether instantiated on grey matter or silicon, there’s more to a mind than the number of neurons. It’s possible we’ll stumble our way into AGI by throwing more hardware at the problem. But I’d like to think that if we ever get there, it’ll be thanks to having cracked the code of our own minds.
1 – In the brief research I did for this post, I found 0.28mg as the mass of a Cataglyphis bicolor ant brain and 9 kg as the mass of some of the largest sperm whale brains. According to those numbers, the whale brain is 30,000,000 times bigger than an ant’s, adding a 300 factor to the already impressive 100,000 times difference. I chose to report Kelly’s data because his work inspired this article. Five zeros make as compelling an argument as seven.
2 – There are more refined measurements still. The encephalization quotient, EQ, is “a relative brain size measure that is defined as the ratio between observed and predicted brain mass for an animal of a given size, based on nonlinear regression on a range of reference species.” Humans fair better using this approach, but the resulting ratios against other animals don’t seem to map with everyday experience. Humans have EQ 7.6. Dogs 1.2, only six times less. Or maybe there’s something there… After all, dogs get us to house them, feed them, and take them to the park, all for free.