Speaker 0:  we've spent decades trying to make silicon think like a brain, but cortical labs just bypassed the middle man by teaching living human neurons to play doom. if you care about massive power grid crisis strangling the ai industry, the 3000% leap in biocomputing efficiency just changed the entire timeline. yeah. living brain cells are playing doom now. this is what they're doing. cortical labs successfully trained biological computer chip containing 200,000 living human neurons to play the complex video game doom in less than a week. by placing lab grown neurons on a microelectrode fractional of the fractional of the fraction of the energy of traditional ai, creating a legitimate pathway for sustainable computing and advanced robotics. what do you think about this, john?

Speaker 1:  yeah. i think this is really cool. that, i love that doom is used the, as kind of the benchmark for a lot of things. so, like, people have put doom on calculators. they put doom on pretty much any electronic that you can think of. that's not this, but i just like that doom is this. so, the thing is that the neurons themselves, they're not seeing doom, like, as we see it. they're receiving structured electrical signals, which represent the game environment, and then they're responding through firing patterns. so it's this is really fundamentally different from digital ai because digital ai processes symbolic or numerical data. so this system, it processes bioelectrical signals in real time. so, there's a learning mechanism to this and this is why this is huge. so the neurons are exhibiting adaptive learning behavior, meaning they improve performance over time, they respond to feedback loops, and they develop goal oriented behavior, which is similar to reinforcement learning. but instead of software based neural networks, this is actual biological plasticity. and so you have synapses that are strengthening or weakening based on activity. so there were early experiments. i don't know if we covered it, but it taught neurons to play palm. but doom is far more complex because there's three d navigation, there's real time enemies, and there's multi variable decision making. so the sexy the success here in what they've done, it shows that biological systems can handle higher order cognitive tasks and not just simple reflex loops. but this sounds cool. there are some performance and limitations. so the neurons play like a complete beginner. so think of your mom or dad playing doom, trying to play doom. the learning is slow and it's actually unstable. and the system is still extremely primitive compared to a human brain, obviously. but this milestone is not the performance. it's more of the proof of concept. and so i think this is cool. this is, this is a huge breakthrough. what are your thoughts on this, craig?

Speaker 0:  i think it's an amazing breakthrough. and not only, did they be able to disapprove upon what's going on with pong, but they are able to the doom demonstration used only 200,000 neurons, roughly a quarter of the neurons used previously in pong. so we're just able to do much more complex things at a quarter of the neuron use. and with a learning efficiency over 3000% over historical, uses of this biological type of system. though, this proves biological systems can handle real time adaptive decision making at a fraction of the energy cost of traditional ai, creating a legitimate pathway for sustainable computing and advanced robotics. i think of this, like trading a massive fuel guzzling freight train for a highly trained homing pigeon. traditional ai models require immense power and massive data centers just to brute force a route, but this biological chip uses native built in instinct to adapt and find the target on its own running on barely a fraction of a watt. so from a energy perspective, this sounds amazing, but i do have part of me that says, is this the first step towards the bork?

Speaker 1:  maybe.

Speaker 0:  you do wish to be assimilated.

Speaker 1:  yeah. yeah. is this like, where does this go? i know we've covered, brain chips too as well in the past, and this is similar to that. but, i think that the biggest break the hidden breakthrough here, like you talked about, is the energy efficiency. so these systems, they can outperform gpus by orders of magnitude and energy per computation. so, maybe this is elon's way of getting, of getting those data centers up in space, but, or whoever's getting them up there. but, i think that the cool thing about this is that they didn't program it. right? they didn't program it to play doom. they said learn it. so it learned organically, and it's it's unpredictable. just it's like a brain learning. right? so this is essentially challenging the entire software paradigm. so like you said, are we assimilating like the board? there are some ethical implications, and they're massive. so the question is, are these neurons aware at any level? can they feel pain or stress? and where do we draw the line between tool and organism, borg? so the researchers are emphasizing that the systems are not conscious, but the trajectory raises some serious concerns. and so some, yeah, i'm concerned about this, but i also think it's cool.

Speaker 0:  so researchers and developers can actually purchase the c l one it launched, a little while ago, and it cost 35,000 us, unit. and developers can interact with the living system using standard cloud based python api. so this is kinda interesting. the neuron count per unit is about 200,000 human neuron neurons per c l one unit. the energy efficiency is estimated a million times greater energy efficient than modern silicon based systems. so it almost feels inevitable. like, if this really is a million times greater efficiency, it's almost inevitable for us to start using biologics to, as chips.

Speaker 1:  yeah.

Speaker 0:  so this operates on the organoid intelligence. the c l one chip uses a microelectrode array for two way electrical communication. the network doesn't rely on static feed forward algorithms. instead, the living neurons undergo genuine biological adaptation through membrane potential changes and synaptic weight adjustments in response to stimuli. so this is the most radical shift in processing architecture since the transition from vacuum tubes to solid state transistors, moving us from artificial, neural networks back to literal biological ones. and like i said, we're heading into a power crisis because we want all these data centers that, you know, cost so much in energy. and if this is, literally a millionth of the energy cost, this has to be being looked at as a solution.

Speaker 1:  yeah. just sent you a link in the private chat to the picture of the c o one. i i was picturing i don't know what you were picturing, but i was picturing something completely different when it comes to what we were talking. i thought it was gonna be very rudimentary, but this seems pretty advanced, which is kinda cool. it's, like, ready for commercial use.

Speaker 0:  yeah. the website looks good and, the device looks interesting.

Speaker 1:  yeah. i know. right?

Speaker 0:  and you know, again, we're definitely interested in where this goes. while these in vitro neurons excel at rapid adaptation, they currently lack complex neurochemical dopamine reward pathways that human brains rely on for long term memory retention. so we're we'll have to see where that goes. the pursuit of neuromorphic computing is heavily funded by legacy tech giants like intel as its low high chip and ibm has true north, but both attempting to mimic brain architecture and silicon. the cortical labs is leapfrogging them by simply using raw biological material instead. the question becomes, though, we're talking about we saw that unit. it's not small. so we have to really take a look at size per computational power even if it is a lot cost efficient. do we need, you know, 10 times the space? we're not talking about nanometer, four nanometer technology like they have today. so we'll have to keep an eye on this. the original 1993 release of doom is the ultimate computing hardware benchmark. as you were mentioning, it's been notoriously ported to everything from atm machines to pregnancy tests. apparently, it can run doom on a pregnancy test. and now holds

Speaker 1:  that word. yeah.

Speaker 0:  complex virtual environment navigated by disembodied human neurons. so we'll have to see what happens because the system currently long lacks long term memory capabilities and keeping biological material alive, stable, and sterile at massive commercial scale is a monumental logistical hurdle right now that standard data centers are not equipped for. so keep an eye on what's going on in singapore. if one data centers can prove stable long term uptime across a thousand biological units, biological cloud computing becomes a viable, enterprise reality. so biocomputing just crossed the threshold from fascinating science experiment to scalable infrastructure offering a genuine genuine off ramp from ai energy crisis. would you trust a living neural network to run your smart home, or does a biological computer cross a creepy line? let us know in the chat. is this where the board starts? what do you think? we'd like to hear from you.