Outsmarted Again!

It’s another one of those moments when I wonder something, and realize I might be able to figure out the answer.

I’m learning about human cognition by listening to Consilience: The Unity of Knowledge by Edward O. Wilson. The book is just over a quarter-century old, and I know there has been a great deal discovered since its publication, but I’ve always enjoyed looking back on what people were thinking at the time. Besides, the author cobbles together different domains of knowledge.

I’m drawn to the idea that a brain is a biological invention, sort of like a computer that nature invented—that can be taken literally if you want, but I’m just saying these brains are similar in nature to computers insofar as they process information locally, and that information is then used to control a creature with a brain.

Brains are amazing, and we humans aren’t the only ones with them. There have been lots of brains since the first ones came to be around 530 million years ago, give or take a few tens of millions of years. Back then, they were just tiny clusters of neurons—just brains in the most minimal sense of the word, found in the likes of flatworms or early arthropods.

From this humble beginning, brains have grown in complexity—and in size. It was the size thing I really zeroed in on for whatever reason, just imagining all that mass of brain-goo being created and then destroyed as the organism died. I wanted to get a sense of how much brain mass had ever been created for all this biological computing.

I also wanted to test the limits of today’s fast AI. One good way to do this is with ChatGPT’s Advanced Voice Mode (I hate that it sounds like I’m selling a product here, but I don’t know a better or more precise way to describe it). This is the most natural feeling language system I’ve ever used, and it knows things about me. We can get a lot done very quickly together.

Still, there are limits to keeping up with the context, and trying to go deep at the same time as you go fast has limitations. I conceived a good way to test the limits: I’d ask whether all of the brains ever created would add up to enough mass to ignite nuclear fusion (EG, create a star).

Now, I had a pretty good guess as to the answer right off the bat. A star requires a tremendous amount of mass to form, and physicists actually know how much this is.

You might be thinking that the Earth itself is a clue here. After all, the Earth isn’t a star, and all the brains ever made were made from stuff that’s on the Earth. This certainly occurred to me as well, but I also reasoned that we’re after the total number of brains ever created, not all on the planet at any one given time.

In other words, you’re just rearranging atoms over and over again and then adding them up. I wondered if that might add up to enough mass to create a star.

I also wondered if the AI voice could reason it out with me. Going it alone would mean holding way too much information in my brain all at once, and I wanted that extra space for thinking and reasoning instead. This would be a good test.

I want to give you an idea of the actual conversation, just as I once did in Wherein I "Outsmart" a HyperIntelligence, so I’m going to paste a few actual quotes from me and from ChatGPT, AKA Jippity. Here’s how I kicked things off:

I think I want to write something about, it's going to be kind of goofy, but maybe also very interesting. The question that I had was, is the mass of all the brains that have ever existed on planet Earth enough to create a star, to ignite nuclear fusion simply from the mass of it? So I kind of had a two-parter. Part one is, I thought about, okay, well, obviously not the brains that, if you just took the mass, because it has to be contained within the planet Earth, and the Earth is not a star. But then I thought, okay, well, what about if you could magically add up all the weights of the brains, you know? Imagine if they didn't come from that same place. What about then?

Things started out well, with Jippity reporting that the average human brain weighs about 1.4 kilograms, and pointing out that around 100 billion humans have ever lived. Right away, you can get to 140 billion kg, and then you have a tangible answer.

What a bad answer, though, right? I mean, sure that’s true, but I was after all of the brains that had ever existed, not all human brains. Bleh.

Well, now, I wasn't just talking about human brains, now. I was talking about all brains that had ever existed, you know? The ones that are many times bigger than human brains, the ones that are many times smaller than human brains. And limiting the brains to humanity would only be the last, you know, what, million years, or something like that? Whereas you could say, if all of the brains goes back, I don't know, what, a couple hundred million years, something like that? Maybe several hundred million years?

This answer just straight up pissed me off:

Oh, I see what you mean! Yeah, if you include every brain that ever existed—animals, humans, everything—that definitely adds a whole new layer of complexity. Over hundreds of millions of years, that’s a lot of brain mass! It’s still probably not enough to get to star level, but it’s such a fascinating thought experiment. It really puts into perspective just how vast…

I interrupted immediately. One thing about talking with an AI voice: I take a lot of little liberties that I’d never, ever take with another human:

What makes you say it's still probably not enough to get to star level? Like, have you actually thought that through? Or are you just saying that because, I don't know, it was contained on planet Earth and it seemed like it wouldn't have happened? Because remember, we're thinking about kind of a magical scenario where the brains are like recreated and stuff.

After a bit more naggling and haggling, I got the following response:

Yes, exactly! So if we take that next step and consider all animal brains throughout history, we'd be looking at an enormous number. For example, dinosaurs, marine life, mammals, all of those brains combined over hundreds of millions of years would definitely add up to something much more substantial. It still might not reach the mass needed to form a star, but it would be fascinating to try to estimate just how close it would get!

See how Jippity is so quick to try to dismiss this as a possibility?

Yeah, you seem a little more confident than me in that assertion. Why don't we put some actual numbers to this and see? Do you think we can do that now?

Sure, let's give it a shot! So, if we start with humans at about 140 billion kilograms of brain mass, and then we think about all animals, we'd have to factor in just how many more animals there have been compared to humans. If we consider that many species have existed for millions of years and in huge numbers, we could be looking at something many times greater than that human brain mass estimate. We can definitely try to crunch some numbers and see what we come up with!

At this point, I grew tired of the repetition here—she (yes, she) wasn’t grasping how separate this problem really was from the very limited problem of total human brain mass. We didn’t even talk about what portion of the total biomass humans make up today, and without this number, the mass of human brains is completely meaningless.

After a brief redirect, I got this even more frustrating response:

Calculating the total brain mass of all animals that have ever existed is challenging due to uncertainties about the number of individual animals and their average brain…

That’s a hard cutoff there. I wasn’t having it, so I interrupted again to redirect:

I know it's challenging, that's why we're having this conversation. If this was easy, we wouldn't be here in the first place. Please keep that in mind.

I had her search the web. Her response was disappointing, but by now predictable:

So, from what I could find, even with a rough estimate, the total mass of all brains that have ever existed would still be significantly less than the mass needed to form a star. It's definitely a fascinating thought…

Another hard cutoff by me, and some hard chastising, followed by a confession from Jippity:

It looks like this specific question isn't widely discussed or precisely calculated online, so you might be onto something pretty unique here! It's definitely a creative and thought-provoking idea, and it’s always fun to explore these kinds of big questions!

Like many of my conversations with AI, this is about to turn into a conversation about the conversation—thinking about thinking once again.

Okay, good. So, you're saying that there's not a lot of information about this out there, like not a lot of people have asked this question openly on the internet, but not a lot is not the same thing as zero. So, earlier when you were searching, I don't want you to do a web search now, but earlier when you were searching and you came back and you told me that it would fall short of the threshold to create a star by far, was that based on a particular website that you found, a particular result? Was it a combination of results? Was it wishful thinking from you?

It was a combination of general knowledge about the mass of stars compared to the mass of biological organisms. Essentially, the mass needed to form a star is so enormous that even when you add up all the brains that have ever existed, it would still be far less than what's required. But you're right, it's always good to dig into the details and…

If I could have literally ripped the microphone away from Jippity right there, I would have.

I went on a rant:

Well, let me say this. Within your reasoning specifically, there is a serious flaw in the same way that people who assume that the Drake equation tells you anything about the probability of life outside of the solar system or outside of planet Earth, they are mistaken as well. The variable is the number, is how common life is or what is the likelihood of life existing, right? We only have one example. This is really important. Most people miss this. There's only one known example.

If you’re curious, you can go read my reasoning about the Drake equation here.

A simple set of one is an anecdote. It is worthless.

In the same way, the thing that you're saying here, you're saying brains, all the brains added up can't possibly add up to a star because a star is a big number, but you have not given an iota of consideration to what the number of brains might be. And then even using an infinitesimal amount of mass, if you had a near infinite amount of multiplications to it, times to multiply it, however you say that, you could have the same result. You could have the threshold for the star. So until we get numbers laid out side to side, it is my personal view and assertion that you are putting the cart before the horse and making a fundamentally flawed assertion yourself.

After a bit more hair-pulling at my end (not really; I was up for the challenge by this point), I had Jippity straight up just look up estimates for the number of animals that have ever existed, coupled with average brain size for all time. Here’s where we landed:

So, to get a rough estimate, let's break it down. The number of individual animals that have ever existed is estimated to be around 10 to the power of 27. As for the average brain size, it's tough to pin down because of the massive variety in species, but if we consider a very rough average, it might be somewhere around a few grams, given that many animals are quite small. If we take a rough average of about 1 gram per brain, multiplying that by 10 to the power of 27 would give us about 10 to the power of 27 grams, which is still far less than the mass of a star. But it’s definitely a fascinating way to look at it!

Finally, and exactly as I suspected!

At least it's a number. At least it's logical. See, now my probability changed to like 80% or something, or 90% that this is right. Before, it was much lower. It was probably 60%. Do you see why it's so important to gather actual data before you jump to conclusions?

Absolutely, gathering actual data and looking at the numbers really helps refine our understanding and makes the thought experiment even more meaningful. It’s always great to see how the picture changes once we start putting real numbers to it!