Breeding octopuses

Our last common ancestor with cephalopods dates back 600 million years. For reference, dinosaurs roamed Earth only 252–66 million years ago.

This ancestor was likely a small, flattened, worm-like creature with primitive eyes or light-sensitive patches. At the time, no organisms had made it onto land, and the largest animals were likely sponges and jellyfish. This evolutionary split is nearly twice as old as the one between mammals and birds. One path led to vertebrates, including us. The other led to invertebrates like mollusks, crabs, bees—and octopuses.

Cephalopods are a separate evolutionary experiment in complex cognition. Our last common ancestor was so primitive that we can say evolution built intelligence twice.

Octopuses display mischievous, problem-solving behavior. Aristotle noted it. Divers experience it. In experiments, octopuses open jars, navigate mazes, and escape enclosures in ways that suggest true problem-solving ability.

Their nervous system is unique—60% of their neurons are in their arms. Their limbs act semi-autonomously, exploring the environment until the brain "takes over" and directs them more consciously. This mix of centralized and decentralized control makes them unlike any other intelligent creature.

Octopuses are, in many ways, the closest thing we have to an alien mind on this planet.

Medawar and Williams effects explain why evolution doesn’t favor longevity:

Medawar’s Hypothesis: Natural selection weakens with age. Since most organisms don’t survive long in nature, harmful late-life mutations accumulate, leading to rapid aging.
Williams’ Hypothesis: Traits that boost early survival and reproduction persist—even if they cause rapid decline later (antagonistic pleiotropy).

Octopuses take this to an extreme: they reproduce once and then self-destruct. Their optic gland triggers a hormonal cascade leading to starvation, tissue breakdown, and even self-mutilation—a programmed death to clear space for offspring.

Most intelligent animals—elephants, primates, corvids—live long enough to learn, adapt, and even pass on knowledge. Octopuses don’t. Their intelligence emerges in a vacuum, with no time to accumulate experience or culture.

If we selectively breed octopuses to bypass programmed senescence, we could:

Observe whether extended life amplifies intelligence
Improve our understanding of complex cognition
Give these remarkable creatures the dignity they deserve

>>2524
Octopi are good evidence of what I call "the worm mind hypothesis": full general intelligence emerges VERY early in evolutionary history, with worms or earlier. All later elaborations are just scale of different types of compute, and various instincts that aren't all that important to the nature of the mind. There is nothing special about humans or language, except that we have big brains, a lifestyle to make us of them, and language to pass on our ideas.

You mention primitive flatworms as a probable common ancestor between vertebrates and molluscs. I have a jar filled with flatworms (planaria) and observing them a lot dispels any notion that they aren't basically intelligent. They navigate around, find their way through and around obstacles, hunt for and take down prey, articulate their bodies competently in ways modern software still can't match, avoid threats, and probably learn. So I wouldn't say evolution invented intelligence multiple times. Rather that it emerged very early and is actually very basic to the fabric of life.

If you watch video of single celled organisms engaged in predatory behavior, you will see a similar intelligence. No they don't have much sophistication or compute, but the essential patterns of will, knowledge, learning, etc are there. Intelligence is evolutionarily early, at least a billion years old.

I think much AI research in the Altman paradigm (scale is all you need) is missing this. First you need the worm mind. *Then* scale is all you need.

>>2526
>So I wouldn't say evolution invented intelligence multiple times. Rather that it emerged very early and is actually very basic to the fabric of life.

Intellience does seem to be a rather general characteristic of life itself. I was perhaps rather hasty, I meant to say that complex intelligence capable of self-reflexivity (or something of that sort) emerged later on... It seems that as streams of information come in from different senses, they are brought together into a single picture, a unified scene. This subjectivity (which doesn't necessarily entail consciousness, at least in its fullest sense) seems to emerge at greater levels of complexity and allows for harnessing that ''natural'' intelligence even more.

Consider, for example, an organism referred to as "dicty" by its fans: these unicellular social amoebas self-organize, and even engage in farming. In certain conditions, they will sacrifice themselves to build a "space ship" populated by a group of colonist amoebas in the center. Most of the time, however, dicties are quite libertarian. They become social in emergencies. Not unlike humans:
https://en.wikipedia.org/wiki/Dictyostelium_discoideum

>>2546
Amazing. Zero neurons and still closer to AGI than OpenAI. It seems totally obvious to me that intelligence is about architecture and not about or scaling or data or FLOPS or any of that. Intelligence is the open-ended nonlinear cybernetic feedback loop of adaptive forward-looking engagement with reality. It can be done at the scale of a single-celled organism, or the scale of a jupiter-brain.

>>2551
It's self-organization that's the magic of life, not reason or intelligence.

For example, you beautifully wrote:
> Intelligence is the open-ended nonlinear cybernetic feedback loop of adaptive forward-looking engagement with reality.

Let me try to rephrase: Life is the open-ended nonlinear cybernetic feedback loop of adaptive forward-looking engagement with reality.

Intelligence helps, especially as stakes become higher and systems larger, but it's life that's the product, and intelligence that's a feature.