What's weird is that so many people shrug this off with "eh, it's what they said about the calculator".
Which to me is roughly as bad a take as "LLMs are just fancy auto-complete" was.
I feel it's worth reminding ourselves that evolution on the planet has rarely opted for human-level intelligence and that we possess it might just be a quirk we shouldn't take for granted; it may well be that we could accidentally habituate and eventually breed outselves dumber and subsist fine (perhaps in different numbers), never realizing what we willingly gave up.
Our thumbs, ..., our intellect, and especially language, gave us an ecological/economic niche.
We became a technological species.
We observed, standardized and mechanized our environments to work for us. That is our niche.
But then things snowballed in the last couple of centuries. A threshold was crossed. Our technology became our environment, and we began adapting the environment for our technologies direct benefit, for own indirect benefit.
Simple roads for us at first, then paved for mechanized contraptions. Wires for talking at first, then optimized for computers. We are now almost completely building out a technological world for the convenience and efficiency of the technology.
And once our technology frees us from dependence on others, a second threshold will be crossed. Then neither others or the technology, will need us.
I don't see a species of devolving humans, no longer needed by their creations, in a world now convenient for those creations, finding a happy niche.
If there is a happy landing, it will need to take a different route than that.
It seems like a stretch to argue that we have any clue what the evolutionary consequences would be for something that's been around only a couple of years. Human-level intelligence took millions of years to evolve even when the lifespans of our ancestors were shorter than they are now, so trying to predict how something so new will affect the biology of future generations seems like it would be pretty much impossible to reason about. Even trying to predict how technology will affect society in a single generation is hard enough, and that's hardly long enough for any noticeable evolutionary changes to our intelligence as a species to become noticeable.
I don't know or really care what other people are doing with LLMs.
I have learned so much the past 2.5 years it is almost hard to believe.
To say I am getting dumber is just completely preposterous.
Maybe this would be leading me astray if I had the intelligence of Paul Dirac and I wasn't fully applying my intelligence. The problem is I don't have anything like the intelligence of Paul Dirac.
People who make that retort forget that the calculator was immensely helpful but _also_ antiquated the need for mental math, which in my opinion is a bad thing. (Everyone should be able to calculate 5% and 10% of numbers, given how easy it is to do)
Well, I suppose many people do not know of these "mental math tricks".
To get 10% of a number, just move the decimal left: 10% of 40 -> 4.0.
To get 5% of this number, get 10% first, then halve it. It is the half of 10%, which in this case would be 2.0.
If you want to do this on a computer / calculator, you simply do: 40 * 0.10 for 10% or 0.05 for 5%. I was a very young kid when I learned to do this on a calculator, and I absolutely loved it!
Exactly, but when you learn math by calculator, you don't learn these tricks, which brings us to today, where tipping calculators rank pretty well on app stores.
Which to me is roughly as bad a take as "LLMs are just fancy auto-complete" was.
I feel it's worth reminding ourselves that evolution on the planet has rarely opted for human-level intelligence and that we possess it might just be a quirk we shouldn't take for granted; it may well be that we could accidentally habituate and eventually breed outselves dumber and subsist fine (perhaps in different numbers), never realizing what we willingly gave up.