Darwin would have a lot of things to say about a hypothetical creature that was utterly dependent on other creatures to feed it and allow it to reproduce, and few of them would be positive.
All sexual reproduction depends on others.
I assume that if you magically took Darwin to the present age, after you explained what a computer was and what AI was, he'd probably say "huh, I have no idea, this is all so far beyond everything I did in the mid 1800s that I have no idea if the analogy holds or not… by the way, have any of you figured out the mechanism nature uses for storing the information passed on via inheritance? Only we didn't have much of a clue in my time."
Also: all the other stuff you listed is still dependent on another entity.
Life is dependent on either other life or the sun. But a plant can naturally grow its own 'solar panels' and then make more plants. ChatGPT can generate a picture that tells some autonomous agent how to build a solar panel.
Even if I just say "yes it's a tool" (some AI clearly are regardless of whether they all are or any demonstrations of them being able to run economically interesting tasks without intervention), not clear how what you're saying connects with the ideas raised in the article?
Your point might just be bias.
Why does this matter? The dependency on others is present regardless. The crow cannot photosynthesise, it ultimately depends on other entities which do. If we made a vat which produced food electrochemically from PV, that a mechanism is involved neither adds to nor subtracts from the crow. If an AI was transcribed from silicon to a bunch of neurons grown in a petri dish — or even by overriding and rewiring an organic crow brain — it neither adds to nor subtracts from the AI.
(If the AI isn't specially trained to keep the crow alive this will end badly, but that is a separate issue, and one that can be ignored because this is all hypothetical).
Darwinian evolution is the process of selection of the fittest over multiple generations where there is some process inducing variations.
The idea and terminology may have originated in biology, but that specific mechanism is one of the standard ways to train the weights and biases in an AI.
They're called, conveniently, evolutionary algorithms.
Dead easy to write, I did one in 30 minutes a decade an a half ago after being annoyed by a young-Earth Baptist creationist preacher in Cambridge city center.
And of course, this is trivially something which ChatGPT can spit out the code for — try it out, see how it works. :)
> Even simple malware has not become independent of coders needing to write new viruses.
They've been doing that for at least long enough for humans born at the same time to be allowed to drink in Texas.
The incremental improvements of the Wright Brothers airplane was 'like' evolution. The design 'evolved' in response to selection for something that would remain airborne in a wind tunnel, and control well without crashing. The Wright Brothers flying machine evolved in a manner similar to life, so presto: it must be exactly like life in all other aspects! We should fear the day when these airplanes take over the puny humans, because they can travel in the sky at hundreds of miles an hour!
You see how stupid that sounds?
>They've been doing that for at least long enough for humans born at the same time to be allowed to drink in Texas.
There exists self modifying malware. This has not reached the level of malware that can reinvent itself to target new machines and new operating systems versions without prompting and design work from the hacker. If it did, that would be far more impressive than what LLMs are doing.
So, like most here, I'm a high IQ human.
Drop me on a random point on the planet, I die quickly. Narrow it down to land, I die quickly. Remove deserts and tundra, I die a little slower in most cases, unless I disturb a tiger, lion, bear, snake, boar, elephant, etc. Put me in a temperate woodland, I can at least make shelter, but I don't know which berries are edible, nor how to prepare wild legumes or tubers, I can't hunt (and I'm vegetarian but hypothetical animals don't count), so I'll probably still die fast. Put me in a random town, I might be able to survive on the charity of others until I can learn the language well enough to be vaguely useful and/or leave, but there's still a big risk of dying from not being able to communicate with local doctors when feeling unwell.
We're all heavily dependent on others; intelligence doesn't make us magically immune to that, neither does the absence of independence make us less smart.
A hypothetical AI likewise can still out-compete us in the economy, even in the cases where it remains utterly dependent on us.
It might (probably will) even know the Lotka-Volterra equations, and how much of an analogy they are — obligate carnivores adapted to only a few prey can wipe out their food source and die off in turn, something similar but the economics of having humans willing and able to power the AI — and yet still proceed.
But that's not what I was talking about, and they have been used to make other Transformer models, which both aren't what I was talking about and not (that I'm aware of) a virus, and also indeed as you say aren't very useful.
…
I think you and I may just straight up be taking about unrelated conversational threads.