[Enginerrrd]

My own personal definition of life is the following, somewhat hand-wavy, theoretically weak, though pragmatically strong definition:

Life is anything capable of sloppy self-replication in a sufficiently complex environment.

Viruses are definitely alive. Something like Tierra [1], Avida[2], or modern variants[3] of mutating copying programs come close, giving rise to whole ecosystems of parasites and hosts and defenses, etc. though perhaps they are lacking the "sufficient complexity" necessary so as not to stall out and stop evolving much. Chemistry provides such a massive environment of complexity that it's hard to replicate elsewhere, though I'd argue it's hardly impossible.

The problem with this article's definition is that, with near certainty, the first progenitor lifeform on earth did NOT utilize ANY of the DNA/RNA protein translation machinery they state are necessary to meet their definition of alive. You can find some surpringly good statistical analyses of this assertion in an unlikely ally: creationist statistical arguments. They prove pretty definitively that life didn't begin with a transcriptase protein popping into existence.

What are the problems with my definition:

The big theoretical hole: You could concoct some hypothetical scenario where something I'd definitely agree is alive replicates using some star trek technology that doesn't allow for the "sloppy" part. W/e, I consider this pragmatically irrelevant.

The second big objection: It allows us to consider many things as "alive" that most would say are not. In my opinion, this is actually a major strength.

Ok, but what about things like crystals? Personally, I consider some type of self-catalysing crystal or quasicrystal to be one of the most plausible forms of original life on earth. That said, they seem to be lacking the "sufficient complexity" aspect. However, are they really? I'm not so sure. Is it possible for a particular pattern of crystal defects to bring about a replication of sorts of that defect pattern elsewhere in the crystal during crystallization? That may be enough complexity particularly when you consider substitutions (when one element in the crystal lattice gets swapped for another). If some type of pattern of crystal defects or quasicrystal or something developed sufficient self-catalysing ability to bring about a form of sloppy self-replication, it's likely that's enough for life provided that the environment allows it under entropy considerations.

[1]https://en.wikipedia.org/wiki/Tierra_(computer_simulation) [2]https://en.wikipedia.org/wiki/Avida [3]https://en.wikipedia.org/wiki/Artificial_life

[smilliken]

I think it's desireable for there to be a zoo of definitions of life. Life isn't a precise mathematical concept, but a family of related concepts.

Self-replication is almost essential, but it's not hard to imagine a future AI that is highly dynamic, intelligent, etc, but does not create copies of itself.

Metabolism is pretty universal. I'd define it as a process that consumes free energy. Arguably viruses don't satisfy this definition. Computers do satisfy it. Most probably don't consider the sun to be alive though, despite it consuming vast amounts of free energy.

Minimizing internal entropy is an interesting one because it's not objective, it requires a model to define entropy. Something can look to us to be high and increasing entropy because we're using the wrong model to understand it. Viruses are static though, so they are definitely not decreasing internal entropy.

[Gunax]

Counterpoint: If viruses are alive then anything can be alive, given that some machine exists which would make copies of it.

ie. if there was a shower-caddy copying machine, then suddenly shower-caddies could be alive.

In a world with a different life architecture, viruses would likely just be random an uninteresting molecules.

[Enginerrrd]

>Counterpoint: If viruses are alive then anything can be alive, given that some machine exists which would make copies of it.

>ie. if there was a shower-caddy copying machine, then suddenly shower-caddies could be alive.

If the shower caddys underwent sloppy replication allowing for sufficient complexity to have ongoing evolution of their own copying mechanism, I see absolutely no problem with this at all. They will rapidly evolve to become much more conventionally life-like.

I find the notion that viruses aren't alive to be totally absurd. They replicate themselves, evolve continuously into an ever-branching tree of different niches, etc.

Is your objection that they don't have their own copying machinery?

Should mistle-toe not count as alive even though it depends on a tree to live? Hell, it appears it's even missing most of the genes necessary for it's mitochondria to produce ATP[1] so if it's somehow hijacking ATP production from the host tree, it clearly has no ability to reproduce on it's own. What about a human male? They have no replication ability without a female host "machine" to make copies of themselves. Are they therefore not alive? I think your objection rapidly descends into absurdity if you closely examine your definitions.

> In a world with a different life architecture, viruses would likely just be random an uninteresting molecules.

In such a world, a virus would not exist. In the same notion, in a world without sources of sugars, starches, and proteins, you would not exist. If you happened to blink into existence in this same hypothetical as a virus, you would be an equally uninteresting collection of molecules that would rapidly disappear. [1]https://www.quantamagazine.org/the-mystery-of-mistletoes-mis...

[Gunax]

No, it's not that they don't have their own machinery per se. It's that their life-ness depends on the context they are found in.

So on Earth, shower-caddys are non-living because there is not a machine which replicates them.

But in some other galaxy there could be a planet where organisms exist in which shower caddys can hijack their reproduction to produce more shower caddys.

[knome]

I think the easiest definition for life, as we know it on our our planet, is to take Dawkins' perspective from 'The Selfish Gene'.

Life is self-replicant genes and the various machines they build around themselves.

No more. No less.

Bacteria are the quintessential gene replication machine. Viruses make complete sense here, hijacking the machines of the bacteria to replicate their genes. The virus is definitely alive. It replicates just fine. It just uses machinery in its environment, found in other cells, to do this, rather than packing it around itself. If life is the genes, there's no paradox at all. A hitchhiker is no less a traveler for the lack of their own car.

Of course, when we think of life, we think of the meta-structures we see, multicellular coalitions of cells that specialize and depend on one another. And that's fine. It's what we interact with and what is important in our day to day existence.

The human body has something like 100 trillion bacteria in its gut, with only 10 trillion cells cooperating to be the human.

From worms to fish, to frogs, to lizards, rats, monkeys and humans, we are essentially a protective tube around the payload of cells found in the gut biome we carry around inside us, just as those cells are protective bubbles around the genes inside them.

Scientists continue to be surprised at the effects these multitudinous populations within us have upon us. But it's not really that surprising if you consider things from their point of view. We're spaceships, carrying them around in an artificial environment, purpose grown for them to thrive in. They send out chemical hormones into our blood stream to signal needs, and they and we have been shaped together over time, and to depend on one another. So we respond in a way similar to what was successful with prior generations of genes and the organisms that contain them.

Does that mean they control us? Obviously not in any intelligent sense. They do add to our wants and needs, creating part of our experience of being. Relatedly, fecal transplants have been used to quell intestinal issues, but too have they been looked at in regards to depression and other mental issues.

What then are we, in this cacophony that genetic replication expresses?

We are gestalts.

Our brains are certainly formed by the genes inside us, but those genes are found in many, and do not encode our experiences, nor our memories, nor our decisions on what wants and needs to value and follow, regardless of which feel better in any immediate sense. Genes have no art, nor science to them. They are mindless, yet from them we are. Minds born of untold trillions of that live mindlessly.

I find Dawkins' further considerations on the ideas of the 'meme', being a replicator that exists only as a pattern in our memories and communications, to be fantastic here, and possibly a perfect way of generalizing life itself.

Much of what we call our self is a selection and rejection by preference of ideas we are exposed to from other people. In our genes, we do not find our languages, nor the metaphors with which they express our experiences, designs and desires between each other. We do not find our sports, nor rules, nor borders, nor a hundred other things we consider daily. These things come from without. From other people. We transmit them to each other, and to our children, and they to theirs. These concepts are within us, perhaps define what we see as 'us', but are not any physical sense part of us. Society lives in our collective minds, encoded in ideas, that exist in brains that are collectives of cells all of which contain genes, genes which, alone in all this tumult, replicate nearly flawlessly with every generation. The containers change around them, but all life shares these uncountable perfect little clones within us.

Perhaps life will one day be generalized as any self-similar replicating pattern in space and time, with various plateaus and groups of life for our joyful categorization to file examples into.

But for now, in our situation, genes themselves remain the best and most accurate definition of life that I can imagine

[resu_nimda]

I’ve been intrigued by Artificial Life for a while; I feel like it’s the way forward for GAI. I think the ML approach leaves out a lot of what is required to develop interesting intelligence.

The main challenges I see are:

1) With any given simulation, it’s going to take a long time before something really noteworthy emerges, and until then how do you know you’re on the right track?

2) What is the fitness function? This has always stumped me. How do you create artificial resources and competition and life and death to drive evolution?

[Enginerrrd]

I've given a lot of thought to this over the years too and come to a very similar conclusion. GAI really probably needs to create itself.

I think that intuition was at least partially confirmed by the successes of GAN's. ...not every problem can be solved that way, per se but it's a good starting place.

The big challenge here though IMO is computational! If you're going to use genetic algorithms to create organisms, you have to then let each of those organisms train themselves. It sounds computationally expensive. And the search space is absolutely massive! It will take some really intelligent effort to bring it down to a level that's feasible.

On that note though, the other thing is that I'm convinced that you also need to include for GAI is fundamental structural arrangements as evolvable parameters. You can argue that this is unnecessary mathematically, but IMO, a little bit of intelligent structure can make the same number of neurons a lot more efficient to train. We see this confirmed in some of the more successful deep learning pipelines. Different components of the whole system get specialized training which gives structure to the whole system.

From an artificial life standpoint, I thought about just offering resources based on correct answers to questions, with exponentially greater resources awarded to harder questions. Questions though would need to be generated in situ of course to avoid memorization.