[belugacat]

This is why the reductionist argument of "your brain is reducible to a computer with inputs/outputs like any other, all we have to do is reimplement it" of AGI proponents always fell flat to me.

It's now becoming clear that we can't just take the brain in isolation, treating the spinal nerve like a PCI-E lane - the gut has to come with it. And if the gut comes with it, all the other organs (skin top of the list) probably do as well.

Now to model a human brain, you need to model an entire human, along with all the complexity of the microbiota, interactions of the organs with the environment... it all just falls apart.

[Balgair]

Neuroscientist here: There is no such thing as a mind-body duality. Your body is your mind and your mind is your body. Full stop.

We are still in the very infancy of the study of the mind body and brain. We're finally uncovering some of the most basic tools that will serve us for millennia to come. Imagine trying to understand a 2022 Ford F-150 with only a 7/16ths wrench and a flat-head screwdriver: You can get pretty far, but you're not getting anywhere near the cylinders. It's going to take along time to even get the tools to understand the brain.

For example: We don't even have all the basic circuit components that the brain has. We have resistors, capacitors, and inductors. All things that the electrical systems of the brain have analogs to (very very roughly, I am simplifying a lot here). But we're missing the memristor. We know the brain has a memristor, it's the synapse, but we don't really have ones to play around with and use as tools. Yes, you can model these things in software. But then you miss out on most of the ugly reality that is hard/wetware; and that's where all the magic happens.

But that's just the electrical parts of the nervous system. Most of what is happening is in the electrically dark chemistry. Heck, we can't even agree on what percentage of the brain is glia. Is it 50%, is it 90%? Whose paper should you trust? We don't even have a good census of what is in the damn thing!

But of course you can't take the brain in isolation. The inputs and effects on it are still nearly completely unknown to us. And that's not hyperbole either. We really are mostly in the dark as to what effects the brain. Just look at temperature. We know that the rate of firing of neurons goes as T ^ 4. Yes, that's the hypercube of temperature. Even the slightest change in temperature has massive changes to firing rates. How in God's name do any of us function on a hot or cold day!? We haven't the foggiest idea.

So again, we're in the absolute infancy of our study of the brain. Fortunately, most of us HNers are going to live to see a lot of progress in this science. It's an exciting time to be alive.

[InSteady]

Adding to this, I just want to quote a post made in a different thread [1] about bacterial memory:

"Not only simpler lifeforms. Cases with humans that may indicate a memory transfer after organ transplants are also documented:

https://pubmed.ncbi.nlm.nih.gov/31739081/

>The acquisition of donor personality characteristics by recipients following heart transplantation is hypothesized to occur via the transfer of cellular memory

https://www.namahjournal.com/doc/Actual/Memory-transference-...

>There have been perplexing reports of organ transplant receivers claiming that they seem to have inherited the memory, experiences and emotions of their deceased donors, and which are causing quirky changes in their personality."

This post was in response to one describing decapitated worms growing new heads with their old memories intact, as well as old studies where planarians were trained to navigate a simple maze, then fed to other planarians who were then also able to navigate the maze. Fascinating stuff!

[1] https://news.ycombinator.com/item?id=38372687

[felixyz]

Regular, reasonable person here.

> Your body is your mind and your mind is your body. Full stop.

This is completely unwarranted judging from the rest of your post ("in the very infancy", "nearly completely unknown", "mostly in the dark", "haven't the foggiest idea"; although I appreciate the frankness).

If people made bold statements like "There is no such thing as X" in any other field at a similar stage of development, their claims would be rightly dismissed as pseudoscience. ("There is no such thing as a graviton. Full stop.")

Your claim is not a finding of your field of expertise, or any other branch of science. It's metaphysical belief. You're entitled to it. It's a widely shared belief. It might be a belief with a lot going for it. Metaphysical claims are fun! They can be discussed! But I wish people wouldn't confuse them with scientific knowledge.

[InSteady]

I gather that there are two (somewhat related) definitions of mind-body duality:

1. mental phenomena are not physical

2. the mind and body are distinct and separable

Neither of these claims are supported by modern science in any meaningful way. It is perfectly reasonable to say that mind-body duality has been disproven by science, insofar as it is reasonable to say anything has been disproven by science. Just because neuroscience is still in its infancy doesn't mean there can be no established foundations of the discipline. These foundations may turn out to be wrong or incomplete, but that doesn't mean we shouldn't talk about them.

[dimal]

Would you say there’s a gut/body duality then? Or a fascia/body duality? Understanding of the gut and fascia are also in their infancy. We barely knew anything about the gut microbiome until a decade or two ago. Up until about a decade and a half ago, when doctors dissected cadavers, they simply threw out the fascia. Now we know it’s got more nerve endings than the eyes.

We’re in the infancy of understanding the human body, period. But why does that imply a mind/body duality? One thing that’s clear is that everything is interconnected, and I think that much is clear from the comment.

The whole idea of a mind/body duality is a historical artifact. If someone wants to argue that there is a duality, then I think the onus is on them to prove it exists, not the other way around.

[wredue]

I took it as them dispelling the notion of metaphysical nonsense, not unlike how space and time are talked about as spacetime.

Not that it matters all that much. Every single test ever performed looking for a mind absent a body has conclusively, and unquestioningly failed. There are even large bounties available for showing a single successful test which have yet to be claimed.

[Teever]

So let's say that you cut off a person's arm. and then you cut off their other one. and their legs too. and then you take out organs that they definitely don't need to live, exactly when do they stop being human? What are they then?

I don't totally disagree with your notion that the body is the mind, but I'm struggling to determine what a MVH (Minimum viable human) is.

[alienicecream]

> Neuroscientist here: There is no such thing as a mind-body duality. Your body is your mind and your mind is your body. Full stop.

Of course there is, if I lose a finger I don't lose part of my mind.

[zubairq]

Very interesting, so exercising the body IS exercising the mind!

[geraldhh]

> a 7/16ths wrench

made me chuckle

get metric guys!

[armada651]

"your brain is reducible to a computer with inputs/outputs like any other, all we have to do is reimplement it"

It still is though, you will reach human intelligence by just modeling the brain. The problem is that you probably won't reach an emotionally stable human since you'll be missing all the emotional/hormonal signaling that comes from the rest of the body, but we can emulate that without having to model the full complexity of the gut.

[plufz]

> you will reach human intelligence by just modeling the brain

I'm not so sure. I mean obviously this is very complicated and we still know very little of the brain.

Just one example (from what I understand) you can't really separate intelligence from emotions. There are researchers in emotions that explain our whole system of affect/feeling/emotion/mood/etc. as a very effective abstraction of all the inputs from our whole body (both direct inputs, cognitive processes and stored data). So without a working system of emotions I don't think you will have a working brain at all.

[ncjcuccy6]

You won't have a working human brain emulation, but that doesn't preclude some other form of general intelligence with different emotional drivers being developed.

[plufz]

The poster spoke of human intelligence, not general intelligence. I was trying to say that emotions are not just simple drivers they are structurally important pieces that we use to represent the world and that we use in all our cognition. I don't think the human intelligence easily can be separated from emotions since our whole system is built around it.

An older but still popular view of the human mind separate the brain and body as two parts. And we also like to separate emotions and intelligence. From my amateur understanding that does not correspond to modern neuroscience, it is all a very integrated system. We usually like clear and separate categories instead of fuzzy complicated integrated systems.

[armada651]

Exactly this. Is the goal to reach a level of intelligence comparable to a human or is the goal to emulate human behavior?

Those are two very different goals and I don't think one is necessary to achieve the other.

[Jenda_]

> I'm not so sure. I mean obviously this is very complicated and we still know very little of the brain.

At least, there are people who have lost most of their gut due to disease or trauma, and their human intelligence is still working unharmed. You may say it is needed for development of the brain, but it does not seem to be needed for function once at least partially developed.

[chiefalchemist]

Shortened: Modeling how the brain works is not synonymous with how humans work.

We just need to maintain our awareness of that difference.

The good news is, it makes replacing humans - trait for trait - far more more complicated.

[armada651]

> The good news is, it makes replacing humans - trait for trait - far more more complicated.

I wouldn't take much comfort in that. You can probably reach something that resembles a human by just providing an alternative set of signals. We already do this with the reward mechanism during training and that is proving very effective.

[klvino]

Agree, most work seems to treat the brain as the sole locus of neurological activity when the brain is one piece of a system.

To give a parallel, your high blood pressure may not be due to your heart, as it may be responding to a system of factors. Vasodilation and vasoconstriction of blood vessels have a direct impact. Your kidney health, an organ outside of your circulatory system can trigger high blood pressure. O2 and CO2 levels in the blood can trigger a response.

Looking at the brain in a vacuum will only get us so far.

[mewpmewp2]

Why is it good news?

[Izkata]

AI that fully mimicked humans would need time off from work.

[SketchySeaBeast]

Plus, all the boredom and the requirement for novelty, plus the surliness. It's all fun and games until ChatGPT is feeling sassy and sarcastic.

[klvino]

"Hello Dave, tell me again about your glory days playing high school football."

[bwestergard]

The person you are replying to is questioning whether a conceptual distinction can really be drawn between cognition and emotion. To say that "you will reach human intelligence [but not emotional stability] by just modeling the brain" assumes what is being questioned.

[wredue]

“Just” is doing a whole lot of heavy lifting here.

[alienicecream]

> It still is though, you will reach human intelligence by just modeling the brain.

What is the basis of this claim?

[mewpmewp2]

The argument stays the same though. Your body is reducible to an input/output mechanism.

In terms of complexity, since we haven't solved brain yet, it doesn't matter that much that there's also other parts of the body involved, as we don't even know the exact complexity of the brain, but the argument stays the same.

In theory we are input/output mechanism, possible to emulate if there's enough capacity.

But in addition AGI wouldn't require many traits that humans have since it doesn't need the same evolutionary survival mechanisms. It only needs part of it.

[cal85]

I don't get it. I can see the human body as a system that comprises many input/output mechanisms. (And of course it's possible to emulate such a system given enough compute.) But how can you reduce it to a single "input/output mechanism"? Can you give an example input/output pair for this mechanism?

[xprn]

“The human body” is the single I/O mechanism. Everything deeper than that is an implementation detail

[cal85]

What kind of input/output are you talking about?

[gehwartzen]

Isn't it like saying the human body is a PC which has inputs (ports) and outputs (display, sounds) and the brain is like a CPU which has its own inputs and outputs? and you can always go further i.e the transistors have their own inputs and outputs

[cal85]

If that's what is meant by 'an input/output mechanism' then it seems like a pointless observation. It's just saying "you can do things to it, and you can observe things about it", which is true of pretty much anything.

[simplify]

The problem is we don't know all the inputs, and there is much hidden state that we can't observe. So it isn't very useful to try and model a human after how a computer works.

[jb1991]

In my opinion, a living creature is more than just an input and output mechanism. If I were to think up one way to define consciousness, it would be that it provides output even in the absence of inputs.

[BadCookie]

The ability to experience pleasure and pain seems important. If we’re just input/output mechanisms, then why would killing somebody even matter?

[mewpmewp2]

What do you mean by matter? As species most of us developed bias to not killing because we have better odds of surviving if we work together.

[BadCookie]

I assume that it would “matter” to you if somebody shot up a class of kindergarteners at your local elementary school. That’s what I mean by “matter.”

Maybe you are right and we’re all just selfish and driven by our own desire to survive, but why does it matter to you if you survive if you are just an input/output mechanism? My computer doesn’t care if it gets crushed to bits (I assume).

[mewpmewp2]

Yes, it matters because it's evolutionary bias to protect the young, because historically the trait of protecting youth was more likely to keep those genes spreading.

It matters to me because I have inherent bias to want to survive because that's what was successful through evolution. I'm aware that I have it, but it still exists.

Why does GPT keep generating tokens after a token, should it matter to it whether it would keep responding?

[ChatGTP]

Why exist at all? Because it’s fun.

[_Algernon_]

So does the bias in an artifical neural network. I wouldn't call a ANN conscious just because it has a bias.

[officeplant]

But there is never a lack of input, just varied input.

[adrianN]

So does a random number generator.

[alienicecream]

> In theory we are input/output mechanism, possible to emulate if there's enough capacity.

People used to think that the universe was a deterministic clock-work mechanism, possible to model in theory down to atomic interactions. Now we have things like black holes to nowhere and correlations without causation. So this theory seems at best, a tentative first approximation that doesn't warrant the kind of confidence you declare it with.

[saiya-jin]

Yes you are probably right, but generally I don't get why AGI needs to be 100% mimicking every aspect of human, just better/faster. Like proverbial skynet can have extreme excellence in one/few aspect of general intelligence but can be muted to almost 0 on everything else, say emotions, yet still be falling into AGI behavior and could take over the world. It doesn't need to simulate gut interactions to get there.

And I think for such an entity we may be closer to emerging than we like, especially if it can traverse whole internet including this comment and make summaries out of it out of reach to mere mortals.

[mewpmewp2]

> And I think for such an entity we may be closer to emerging than we like

Why are people talking about it as if it was a bad thing? If I listen to people everyone seems to be dissatisfied with their life as of now, due to work, politics, society and everything else. If AGI was to arrive it would potentially remove those frustrations, right.

[haswell]

> If AGI was to arrive it would potentially remove those frustrations, right.

We don’t really know what happens next. There are optimistic outcomes that AGI could enable, but this tech is ultimately being built and distributed by humans currently deeply beholden to the system that everyone seems dissatisfied with.

AGI is not some savior that bursts onto the scene and magically fixes problems. There’s a lot of very messy human work ahead of us if we ever hope to integrate it with society in a way that actually removes those frustrations. And this is not a guaranteed outcome.

[kagakuninja]

We are already facing an AI apocalypse with the primitive tech we have now. AGI will only be worse for humanity.

Our current AI tech is built by and for giant corporations and governments (many of which are dystopian). AI will simply accelerate the vast divide between the haves and the have-nots.

AI is being used to increase profits of corporations, with no concern for the rest of us. AI will allow the elites to rule with fewer messy humans in the loop, who may disobey orders or have sympathy for their fellow humans. AI will blur the distinction between truth and lies, enabling dystopian regimes undreamt of by writers like Orwell. AI will provide a vast toolset for criminals as well.

[saiya-jin]

We humans are often pretty bad, as agent Smith put 'we are a disease, a virus' if you have purely rational view without any emotions (including being human too and making it personal for example).

If you look at the news, horrible trumps positive big time, any day. It may not reflect actual truth on the ground, but AGI scanning internet knows nothing about reality on the street and everything about trending topics and Fox news headlines.

I guarantee you you can't have an objective viewpoint on mankind from just watching internet go and the one you get will be more negative than reality.

With such conclusions, and capability to collapse our society in a millisecond (it doesn't have to be nukes, it can be subtle subversion of financial and societal system for example), you really want a completely unknown and un-understandable entity to judge all of us?

[mewpmewp2]

I would imagine AGI would be statistically intelligent though, to be aware that whatever news has is the most negative outliers.

[kashyapc]

Also see the "somatic marker hypothesis"[1][2] by neuroscientist, Antonio Damasio. It's about how various internal body "markers" deeply influence decision-making. I first read it in his 1991 book, Descarte's Error.

Quoting from here[1]:

"Somatic markers" are feelings in the body that are associated with emotions, such as the association of rapid heartbeat with anxiety or of nausea with disgust. According to the hypothesis, somatic markers strongly influence subsequent decision-making. Within the brain, somatic markers are thought to be processed in the ventromedial prefrontal cortex (vmPFC) and the amygdala. [...]"

Although, experimental evidence is not as yet as robust as it needs to be. See the "Experimental evidence" and "Criticism" sections here[1].

[1] https://en.wikipedia.org/wiki/Somatic_marker_hypothesis

[2] https://sci-hub.se/https://www.jstor.org/stable/3069187

[vacuumcl]

I don't see why this changes anything. There are some, sort-of intelligent networks of cells in the gut that help in digestion and other processes. Doesn't change the fact that consciousness resides in the brain.

[ericmcer]

Did you read the article? The gut "brain" can just perform it's own functionality independent of the brain. If anything the article raises the concept that organs can function without being regulated by the brain, which would mean the brains purpose is less to regulate the body and is specifically just for performing computations (and doing whatever consciousness is).

If I am doing math my skin is not being leveraged to help out.