Articles, Blog

What self-driving cars can learn from brainless slime mold

What self-driving cars can learn from brainless slime mold

– We know you don’t need one big brain to act smart. Take ants, for example. They build intricate colonies;
they farm; they wage war. Each ant has a tiny little brain. But together, they exhibit
swarm intelligence. That means the group is smart in ways that individual members are not. You can see swarm intelligence at play in other tiny-brained creatures, especially other insects, like bees. But swarm intelligence also appears in a much stranger place,
within a single creature without any brain at all. It’s a little blob
that’s teaching us a lot about what it means to be smart. We took a trip to the swarm lab at the New Jersey Institute of Technology to check it out. (gentle music) – Studying slime mold is, it’s
probably one of the weirdest organisms I’ve worked with. My name is Simon Garnier. I’m an assistant professor
at the New Jersey Institute of Technology. – [Alessandra] Simon runs the swarm lab, which studies intelligence in places you might not expect to find it. – The main research topic of the lab is trying to understand how what we call decentralized systems,
so system that don’t have a boss or an architect
or someone in charge, how these systems are
capable of self organizing and through this self-organization, find solutions to problems. – [Alessandra] And Simon
has an unlikely test subject for his studies. – Slime mold is a unicelled organism, so it’s a single cell, but
it’s a very particular cell, if you compare it to what
people think cells are. Instead of having one
nucleus, it has actually millions of them,
sometimes billions of them. It’s a cell that can grow
over like very large sizes. I mean, one of the rare cells in the world that you can actually
see with your own eyes. – But as unique as slime mold is to study, it also takes a lot of patience. – It doesn’t move very fast,
just a millimeter an hour. For someone studying animal behavior, it’s one of the most frustrating things that you can’t see the behavior directly. You have to wait. Other than that, it’s cheap. It doesn’t taste very good. I wouldn’t recommend people to eat that. – [Alessandra] You tried it? – I tried it. – [Alessandra] What does it taste like? – It’s more like, have you
ever, like, licked the floor? Like, it tastes like dirt. – But slime mold isn’t just
weird because of its size or its taste. It also appears to be intelligent. It doesn’t have a brain
or a nervous system, but it can solve all
kinds of complex problems without any of that hardware. As slime mold grows, it can
keep track of where it’s been. It can solve mazes in search of food. It can even be trained to take risks in the name of a big payoff. And then, there are the
transit experiments. About a decade ago, scientists
at the Hokkaido University in Sapporo had a weird idea. – What the researchers did, essentially, is they gave slime mold a
map of the Tokyo rail system, and each of the station
was actually a food source for slime mold, and
then they let slime mold explore that area, and as I said, when slime mold find
resources in the environment, it’s gonna start building
a network between these different resources. – [Alessandra] What the researchers found a few days later was a pretty
well-designed rail system that closely mimicked the real-life map. – What they found is that the slime molds actually build networks that
are pretty close to optimal. They are very cheap to build, but at the same time,
if there is a disruption in the network, they will
be able to get around it. – [Alessandra] That bit about
being responsive to disruption is something that our
transit networks could learn a lot from. – That’s a problem we
need, like, supercomputer to solve, and then this organism is just essentially a bunch of proteins and lipids and this thing is just capable
of solving it naturally, without any external help. – Simon’s team helped us to recreate another famous transit experiment that uses a map of the United States. By placing food sources
on major US cities, we essentially asked the slime mold to build us an interstate highway system, which begs the same question
that so much of this asks. How is slime mold doing this? Simon’s not entirely sure. But he has an idea. – Here in the lab, one of the hypotheses we are exploring is that
the brain of the slime mold, if you want, is actually the
membrane of the slime mold, ’cause the membrane is
the part of the slime mold that is both in contact
with the environment, with getting the outside information, and in contact with the
physiology of the organism, so getting the inside information. – Meaning even without a brain, a single bead of slime mold
can react to its surroundings and disseminate the information
throughout the cell. Simon thinks this decision-making power may come from that ability
to synchronize information throughout its system. We left our Petri dishes at Simon’s lab for three days to let it grow. And since he mentioned it,
we had to try the stuff. – Cheers. – (mumbles) Cheers. It tastes like dust. – Mm, yeah, it’s like moss. – Yeah. – It tastes like moss. Yeah, it reminds me of
when we used to put moss in our nativity for Christmas. – Oh, and you were eating it? – No, I’m tasting it
instead of smelling it. (laughing) Before we left, we also
asked the big question. What exactly can we
learn from all of this? Well, first, studying slime mold is like studying the
very, very distant past. – Life was unicellular at the beginning, and that’s a unicellular organism. And so, by looking at
how something like this is capable of solving complex problem, we sort of get to the origin
of intelligence on Earth. – And second, he thinks that
those transit experiments aren’t just good metaphors. – In the future, like self-driving cars is going to be a big decentralized system. We’re going to have millions
of these self-driving cars on the road, and essentially
they are all computers on wheels, and each of
these computer on wheels is going to try to make
decision in real time on which road to take, how
to avoid this other car, how to find the best path. – Basically, if we can
tease out the algorithm that slime mold uses to make decisions, we could use some of
that math for ourselves. – If self-driving cars are
capable of talking to each other, figuring out where traffic is clogged, they can automatically
redistribute themselves in the network and decrease, essentially, the amount of pressure on particular road, making it better for everyone. – [Alessandra] A few days
later, Simon sent us the results of the experiment. The map isn’t perfect,
but it is comparable to a highway network that
took millions of dollars and years to create. So yes, this is just a bunch
of goo in a Petri dish, but it’s goo that can make you question your own complexity as a human. – So the question is, if
slime mold is not, say, intelligence, and it’s using
only like basic physics and chemistry to solve this problem, well then, if we are as
good as a sack of puddings, like are we intelligence really, or do we need to sort of really find what intelligence means
when we look at something like slime mold? – Hey everyone! You’re watching this on
the brand new Verge Science YouTube channel. The whole science team
has been working on this, and we’re really excited. We’re putting out a video every week, so please subscribe to this channel and check back next week for more. Thanks.

Tagged , , , , , , , , , , , , , , , , , ,

100 thoughts on “What self-driving cars can learn from brainless slime mold

  1. It's crazy the things we can learn from organism even the simplest of them. Sadly though we are causing an extinction of organisms never seen before.

  2. I'm really into verge lately! They are exploring and sharing some super interesting information! Keep up the great content Verge! 👍

  3. How long does it take for the slime mold to generate these graphs? They are most likely brute-forcing their pathways, and it seems to me that the algorithm it uses is to expand a circle around the areas of interest until they hit another area of interest, at least from the footage at 6:39. That's a pretty slow algorithm in terms of computer science, and would not suffice at all in a real-time situation.

  4. Intelligence is just feedback networks acting in an changing environment, the more complex it is, the more intelligent. Machine learning is still way too simple and has not much feedback with an environment.

  5. well isn't the brain " swarm intelligence too?" it's just a lot of stupid neurons that fire, but the cooporation of billions of neurons result in our intelligence

  6. I love your website and Im happy that you have a youtube channel..I already graduated in college long ago and stop urge to learn more ignites because of your site. I realize that there are too many things to learn and learning doesnt stop as long as we live

  7. + Verge Science The woman is loud enough and very clear spoken to understand. I don't think I one word that the man said. Their voices seem to be set at different volumes. Thank you for enabling captions.

  8. I LOVE slime mounds – I had one in an old house once. They are so cool. One day they are a jelly like blob on a rafter – the next they are 20m away doing a runner. They can navigate. And move.

  9. The thing they should have really shown is that could the slime make the exact same pattern everytime when given the exact same condition. That will test its optimality.

  10. "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius—and a lot of courage to move in the opposite direction." -E. F. Schumacher

  11. sure, It's a relative question. If you ask do you think this is "big"? Big compared to what, a mountain? Is slime molds intelligent compared to a human being, no not really but compared to another lily pad? yes I think it is. But really what intelligent is defined as in these two samples may change the out come, apples and oranges? I don't know.

  12. No matter how big of a motherboard u have,it wouldn't really matter if it doesn't hold SMART PROGRAMMING.

  13. You're welcome, morons. I have done what you could not. I'm not saying I'm better than you, but it is clear that I am.

  14. "Swarm Intellegence" I'm sure the easier term or more accurate is "Emergence"
    Anyways good video, made me more curious about this fascinating slime mold.

  15. Slim mold is absolutely intelligent, as are plants and computers. The question is how smart is it? Smart phones are in a way intelligent, but are they smart than a Squirrel? Probably not. "AI" is a measure of programmed intelligence compared to our biological human intellect, witch is the most advanced we know of. Computer programs could do the same "thinking" as slime mold, they can calculate sunlight and temperature like plants, and perhaps even replicate the intelligence of insects. [(side bar) How similar is an ants brain to ours, or a computers?] Anyways, thanks for the interesting video!

  16. Let me tell u something…. U guys have just got it all wrong!. It is not about (a bunch of proteins as he calls it , that's such a shameful word) it is all about interactions and coordinated connections. The cells collectively don't have intelligence or much of a brain… All what they do is simply talk with each other. Each and every cell grows multiplies mostly in a random fashion and in accordance to any crazy stimuli. And that data that info he sends on to the other nearby cell and that cell to the next cell… By means of which they form a system which is just like a basic version of our body.

  17. problem, this doesn't consider mountains, forests, rivers, etc, you know, obstacles that shaped the road system we currently have

  18. I reckon Trees also communicate and make root connections like that together with fungus underground.

  19. 3:50 why do scientists feel the need to describe things as "just" . why think of such an obviously complex thing as "just" lipids and proteins when those are merely 2 of the many components of it. what do you get from having such a belittling perspective

Leave a Reply

Your email address will not be published. Required fields are marked *