Really cool science, or nightmare fuel? Brain Organoid Computing for Artificial Intelligence

[Nothing Loud]

So as a multi-disciplinary researcher I tend to lurk academic twitter and biorxiv and medrxiv…well this week there was a new preprint (means not yet peer reviewed) paper by schools like UF, Indiana University, Cincinnati, etc that I found positively fascinating but I'm sure many on here may have different views.

https://www.biorxiv.org/content/10.1101/2023.02.28.530502v1.full.pdf

Please read as much of the paper as you can. Does the headline scare you? Does the actual content scare you? How about after reading the whole paper?

I've seen such a variety of responses to AI news that I wonder what plucking an example from the bioengineering biorxiv actually does to resonate with the average (and non-average) era user.

What this paper proposes is a method to build better AI hardware by using artificially-grown human pluripotent stem cells that become advanced synthetic brain tissue that can respond to spatiotemporal electrical pulses to function as a biological neural network (BNN).

It then tests this synthetic living AI hardware on solving math problems that are very hard for current mainstream computing technology to solve quickly (nonlinear dynamics, including a chaotic term), requiring numerical methods to estimate the answer to a math problem, with significant variation due to error. However this "brainoware" hardware can actually handle these types of nonlinear/differential equations more instinctively because of their intrinsically nonlinear physical state.

So you may have not known already that bioengineers do these sorts of things. Frankly it just reminds me that we live in the future already.

The application of tech like this is to demonstrate and propagate advances in AI capability that can solve problems across science and society at a larger scale. But does it come with risks too? What if this kind of technology unlocks the door for more organoid synthetic living computing?

 

[Coothman]

It does seem logical that hardware is a big limitation these days, however, this is a step I wouldn't have imagined we were anywhere near this close to realizing. It's all fun and games until you've accidentally created Mother Brain.

 

[nsilvias]

idk how do you stop the brains from developing a personality and feelings

 

[brawndolicious]

There are a lot of ethical problems with AI that come down to who is using it and for what application but I don't see an issue with building a biological computer that we feed problems into rather than the current grains of sand approach.

I'm guessing what you're asking is what happens if somebody 50 years in the future decides to go to Best Buy and duct tapes a dozen PS13 consoles to make a fully sentient being? I don't know that's an interesting question, I imagine you have to treat that the same as how Child Protective Services treats unfit parents. Maybe beyond a certain level of processing power, you end up subject to a bunch of regulations?

 

[Vulcano's Assistant]

There is two worlds to computers in a way, those that study computing in an on itself and those that do something else unrelated to their making. I can see this existing in labs where they themselves are the experiment, but at scale that drawback of maintenance they mention has to be outweighed by benefits if it is going to exist elsewhere where other methods would have sufficient even if considerably inefficient.

Even still within the world of scientific research. Say you have a special computer running on some biological component solving some specific task more efficiently, but the nature of the hardware is unimportant. You have to weigh the performance gained against the possible drawbacks of it requiring more maintenance or that living hardware could degrade, while silicon chips can be turned on and off, left turned off for as long as necessary because of transportation or installation, etc without risk of damaging it.

 

[Nothing Loud]

idk how do you stop the brains from developing a personality and feelings

Click to shrink...

Depends on the scale and complexity of a brain organoid computer. A small, local tissue of synthetic brain matter is probably incapable of developing personality, but if this type of technology were scaleable to housing warehouses filled with organoid computers, I suppose that is theoretically possible depending on the development of the stem cell architecture.

 

[Nikus]

Ah yes, I've played Metroid.

 

[tejinashi]

Feels like a line is being crossed.

 

[Midramble]

There is two worlds to computers in a way, those that study computing in an on itself and those that do something else unrelated to their making. I can see this existing in labs where they themselves are the experiment, but at scale that drawback of maintenance they mention has to be outweighed by benefits if it is going to exist elsewhere where other methods would have sufficient even if considerably inefficient.

Even still within the world of scientific research. Say you have a special computer running on some biological component solving some specific task more efficiently, but the nature of the hardware is unimportant. You have to weigh the performance gained against the possible drawbacks of it requiring more maintenance or that living hardware could degrade, while silicon chips can be turned on and off, left turned off for as long as necessary because of transportation or installation, etc without risk of damaging it.

Click to shrink...

These were my first thoughts. I don't see how this will ever win out over silicon with the maintenance and degradation hurdles.

I more expect, with our growing tools on protein folding and cellular engineering, that we will go down the route of carbon based "grown" organic semi conductors that'll have design capabilities to create NNs with higher neuron density and parameters than even the human brain. I feel like carbon organic semiconductors has to be the next big leap beyond lithography.

Edit: Little article on the subject. https://www.chemeurope.com/en/news/1176664/flexing-the-power-of-a-conductive-polymer.html

 

[Mochi]

I feel like even if this particular experiment is still outside of the bounds of torture, it leads to future experiments and technology that are morally reprehensible. It makes sense why they would try this, but it still feels like "just because we can, does that mean we should?" tech.

addendum: it's still the same question with AGI, actually, even without the use of organic matter.