▲ 89 points
back
126 comments
For those interested beyond the new age marketing speak that shouldn't exist on any real research company:
These are the guys behind Dishbrain, which is not really a brain but a patch of human induced pluripotent stem cells grown on the Maxone silicon chip [2] which allows recording from the entire ~5x5mm chip with high resolution.
[1] https://newatlas.com/computers/human-brain-chip-ai/
[2] https://www.mxwbio.com/products/maxone-mea-system-microelect...
Ah, sweet. Man-made horrors beyond my comprehension.
Hopefully this will lead to brain implants that can help with solving a bunch of brain issues, and not to what we're all thinking of.
May I ask if you have a mouth, and/or a commitment to screaming?
Never thought the day that brain in a vat is no longer a philosophical question but a practical one would come so soon.
See "MMAcevdeo": https://qntm.org/mmacevedo
One day we'll have biological networks of the size that can compete with today's biggest artificial neural networks, capable of running things like, well, ChatGPT. Then the philosophical questions will run even deeper....
Since they're both neural networks, what questions have changed? I don't see what that clarifies or confounds about the nature of consciousness.
To be honest, i thought it would first go the other way around. To many billionaires afraid of death, trying to encode there mindstate-personality into a machine and become semi-imortal.
And some billionaires aren't afraid enough, of death. Steve Jobs could have outlived his pancreatic cancer if he got it treated on time and didn't indulge in fake cures.
Honsetly if you're an 80yr old billionaire who has done 'everything', why not go for some mind encoding shenanigans?
"Neural networks" as in chatgpt and friends have almost nothing to do with "neural networks" as understood in biology, more than a passing analogy to how synapses fire. I'm very skeptical of projects which seek to conflate the two in some vague way.
This gives me nightmares tbh. Love that they say that they’re happy and healthy, lol.
Very "I Have No Mouth, and I Must Scream" for me.
So head cheese.
Or Bio-Neural Gel Packs.
Depending on your sci-fi of choice.
Don't forget to feed your PC every day otherwise it gets grumpy and will drop all your tables.
The Neurally Controlled Animat: Biological Brains Acting with Simulated Bodies (2001)
This does not appear to be new or novel technology, aside from the fact that the neurons-on-chip are housed in a portable incubator.
Example:
>What happens if we grow a mind native to the infinite possibility space of digital computing?
Nothing we need to care about, as there is no such a thing. Mankind has put a significant portion of its limited attentional power on building interconnected silicon computers, that in a space whose scale is so small compared to human bodies that illusion of infinity is easy to fall into. In the same time, mankind went with global policy of massively draw on non-renewable energy stock, destroying vast sustainable life supporting environment in the process. There is nothing like unlimited resources and infinite space.
Now, obviously, this page is marketing idle talk, with a weak connection to the actual work in their labs.
From a purely scientific point of view, I wonder if that kind of device is just as vulnerable to magnetic storms as a pure silicon based device.
From a human perspective, without much more context, this seems just horrific and I wish them much ethical and legal barriers to stop them already.
If the goal is to mimic the human brain, how is this different than just...starting with a human brain? Either way, you will be subjugating a possibly conscious mind into servitude. I am sure they are "happy" (however you want to measure it) about it either way lol
> you will be subjugating a possibly conscious mind into servitude.
Is the distinction between a protein-and-salt-water model versus an electronic-gates-and-memory model meaningful? From a computational standpoint, they both manipulate states and store data. Does substrate matter unless we're invoking metaphysical claims?
Regarding the video (https://twitter.com/Scobleizer/status/1716312250422796590), the device seems untypically polished but also a bit sus. Even if it contains living neurons, their functionality appears limited to mere survival rather than meaningful data processing. The claim that it's/can-be "more efficient than a GPU" (even if in the future) is premature at this point IMO.
Trying to get this kind of accuracy and bandwidth for signal measurements you'd have to risk killing the human or at least severely cripple them. Look at how outraged people here regularly get over Neuralink's tiny chip that would only insert an inch or so of electrode threads into the brain. Now imagine they were sticking stuff in there that penetrates the entire brain. It's much easier publicity wise to just do that with a bunch of neurons on a board.
The actual paper: https://www.cell.com/neuron/fulltext/S0896-6273(22)00806-6
Of note: This is an active test of the "Free Energy Principle" [1] with seemingly supportive results.
Their definition, in the paper, of sentience is as follows:
> It is proposed that these neural cultures would meet the formal definition of sentience as being “responsive to sensory impressions” through adaptive internal processes.
Their blog is one post per year at https://corticallabs.medium.com/
In the Rifters trilogy by Peter Watts this technology is referred to as “head cheese”. It’s used as a form of AI as it turns out silicon can never beat biology at analog reasoning. It ends up playing a role in the plot.
IRL If the masses of neurons start getting large this starts to be ethically questionable… in direct proportion to how large.
From the paper:
> Under this theory, BNNs [biological neural networks] hold “beliefs” about the state of the world, where learning involves updating these beliefs to minimize their VFE [variational free energy] or actively change the world to make it less. If true, this implies that it should be possible to shape BNN behavior by simply presenting unpredictable feedback following “incorrect” behavior.
Incorrect behavior is punished by pushing static, simulating an "unpredictable" environment, so that the BNN (biological neural network) acts differently. I really hope that if they continue this line of development that they find something other than "chaos overrides" to sensory input because at a certain level of intelligence, that would lead to insanity.
It's clear that you can get some actual neurons in a dish to behave as an artifical neuron network. It is a fascinating concept as a research question. What's not clear is what is the business value here - do they expect these natural networks to outperform ANNs for any foreseeable application?
Yeah, power-performance is a big one. The entire human brain runs at about 12 watts. Human neurons are several orders of magnitude more efficient for neural computing than anything that's been dreamt up in silicon. That might not always be the case, but I imagine there are still another several orders of magnitude at least that artificial neurons might fill as this technology evolves.
Imagine something like a Raspberry Pi with a neural network mounted on it that is significantly more powerful than an h100.
Science does not need a business value.
Surely you'd just want to copy how the neurons work enough to run them on normal silicon rather than dealing with wetware? Or has that hit a wall?
Amazing, scary, and perhaps absurd, all at the same time!
We don't understand how human neurons in particular are so good at what they do. We have some understanding, which has been mathematically approximated with back-propagation algorithms and transformers. That's why we have diffusion nets and LLMs that are as good as they are.
But the nuts and bolts of how neural systems give rise to reasoning or even what consciousness is... we only know that neural systems are responsible for reasoning, and we don't even know definitively that neural systems cause consciousness. We only know that if you suppress their activity consciousness is retarded, and if you damage them enough consciousness is snuffed out.
Beautiful website. They massively over-claim on their science though. Their 'tech' is nothing special, people have been culturing and recording from neurons for years. And their claims of 'sentient' neurons in a dish should be taken as being somewhat flexible with the meaning of the word.
The website reminds me of mid 90s multimedia presentations on CDs. It takes a long time to scroll and read and there is not much to learn about them. Maybe their research page would be a better starting point https://corticallabs.com/research.html
The latest article is "Critical dynamics arise during structured information presentation within embodied in vitro neuronal networks"
When I see something so over-engineered on the aesthetics front, my knee jerk reaction is to think "this is overcompensating, and i guess they don't really have what they promise"
Just finished watching the Pantheon (2022) series about “uploaded intelligence” last week. Pretty ironic timing!
Sadly the show is not to be continued. IMHO very nice show, and riddled with less nonsense and rubbish compared to the website of these guys…
More like, “What a coincidence!”
This is kind of the plot of the third season of Sword Art Online.
No ethical considerations to be found anywhere. Typical.
> Neurons are cultivated inside a nutrient rich solution, supplying them everything they need to be happy and healthy
Neurons are considered organic entities that are satisfied with an healthy environment. It is supposed that nothing suffers.
Our biOS composes their reality, sending information about it via electrical signals. It then converts the neuron's activity into actions inside that reality. Their world is mediated through our biOS.
The simulation argument asserts that "at least one of the following propositions is true: (1) the human species is very likely to go extinct before reaching a “posthuman” stage; (2) any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof); (3) we are almost certainly living in a computer simulation."
(1) is still possible - humankind is armed to the teeth, this tech could be a bunch of hot air, the breakthrough is still N years away, etc.
(2) becomes less likely with every headline. If we had the ability today to simulate a consciousness-filled universe thousands of instances would be spun up overnight.
(3) is looking more plausible than ever...
I know that smarter people than me have discussed this to death, but IMO, you simply can't make statistical arguments like that.
Let's say you have an input x and a step function f such that the sequence x, f(x), f(f(x)), ... contains (in whatever sense) a conscious mind. Once that sequence is defined, it doesn't (and can't) logically matter whether it is evaluated once, twice or a hundred times, whether it is evaluated on a slow computer or fast computer, or, in fact, not evaluated at all. There is always only one sequence, and the act of evaluating adds no information to it.
Eww, what is up with their website? Just scrolling down lags my entire computer.
Bad idea#328 : a fancy website where the client-side script is in fact using the user devices to handle batches of neural networks training units.
Better accessed with PgUp, PgDown
Or, actually, with Ctrl+U (the code is clean, just go to the <main>)
Well, for convenience:
==== ==== ==== ====
###~ What does it mean to grow a mind? ~###
The human mind is the north star for digital intelligence. But silicon can only do so much. Cortical is growing human neurons into silicon. Their reality is our simulation. We think these minds will learn better than any digital model and breathe life into our machines.
###~ Human neural networks raised in a simulation ~###
The neurons exist inside our Biological Intelligence Operating System (biOS). biOS runs the simulation and sends information about their environment, with positive or negative feedback. It interfaces with the neurons directly. As they react, their impulses affect their digital world.
###~ Our first minds ~###
The dishbrain is currently being developed at the CL0 laboratory in Melbourne, AU. We bring these neurons to life, and integrate them into The biOS with a mixture of hard silicon and soft tissue. Our first cohort have learnt to play Pong. They grow, adapt and learn as we do.
###~ Silicon meets neuron ~###
Neurons are cultivated inside a nutrient rich solution, supplying them everything they need to be happy and healthy. Their physical growth is across a silicon chip, which has a set of pins that send electrical impulses into the neural structure, and receive impulses back in return.
###~ A direct connection to infinity ~###
This creates the highest bandwidth connection possible between an organic neural network and a digital world. Our biOS composes their reality, sending information about it via electrical signals. It then converts the neuron's activity into actions inside that reality. Their world is mediated through our biOS.
###~ The Ultimate Learning Machine ~###
Those actions have a positive or negative effect in biOS, which the mind perceives, adapting to improve that feedback. The human neuron is self programming, infinitely flexible, the result of four billion years of evolution. What digital models try and emulate, we begin with.
###~ Why? ~###
There are many advantages to organic-digital intelligence. Lower power costs, more intuition, insight and creativity in our intelligences. But most importantly we are driven by three core questions.
###~ What will we discover if our intelligences train themselves? ~###
We know an organic mind is a better learner than any digital model. It can switch tasks easily, and bring learnings from one task to another. But more important is what we don’t know. What are the limits of a mind connected to infinity? What can it do with data it literally lives in?
###~ What happens if we take a shortcut to generalised intelligence? ~###
Machine Learning algorithms are a poor copy of the way an organic neural network functions. So we’re starting with the neuron, replacing decades of algorithms with millions of years of evolution. What happens as these native intelligences start solving the problems we’d previously left to software?
###~ How can we surpass the limits of silicon? ~###
Silicon is raw, rigid, unchanging. Our organic neural networks sit on top of this raw power, but the way they grow and evolve isn’t limited to the software they run on. There is no software, it's coded in their DNA. How will computing change as we shift from hard silicon to soft tissue?
###~ RFN: Request For Neurons ~###
The dishbrain is learning and growing in biOS today, and soon we’re opening an early access preview for selected developers. The biOS is our simulation environment, where you can program tasks, challenges and objectives for our minds. Join our developer program to get early access to our SDK, and secure training time with our minds.
###~ What comes next ~###
We’re not making smarter computers, more efficient data centers, or more personalised advertising. We’re doing this to see what happens. What happens if we grow a mind native to the infinite possibility space of digital computing? We wonder what it will mean for digital spaces, for robotics, science, personal care. To explore the delineation between the personal mind, the distributed mind, digital and physical realities. To blur those boundaries. We wonder what it means to grow a mind, born of the physical world, but a native of the digital world, where that mind will go, and what it will teach us. Wonder with us.
Yeah it's pretty unusable site. Clearly designed more for marketing than science.