Why Scientists Are Combining Neurons With AI

Ajouté : 2026-05-12

[00:00:00]Biological AI or synthetic AI or maybe you just want to call it intelligence is progressing rapidly. This is where we take biological components and combine them with artificial intelligence. And one independent researcher did publish that they had combined brain organoids of the tiny human brains that you can grow in a literal jar if you want to with a language model and that is pretty neat and we'll talk about it. This research group made a three-dimensional harness that can be combined with brain organoids and it can have continuous communication back and forth and it can coat the entire organoid. This chip can be used in real time to train AI. Yes, AI can be trained on the spiking patterns of our own brains or the tiny brains that we grow. But why? Why would somebody want to do this? Well, there are a few reasons. The one of course is to create artificial intelligence. The spiking patterns in our own brains create logic and that logic can be integrated into an AI. There are more kinds of AIS than LLM, although those can also be combined with LLM. The other reason is that it creates integration for medical devices. We are rapidly approaching the point where strokes can be treated with brain organoids or even modifications of existing brains and they're going to need an interface. That interface could work either with the brain organoid or it could work with our own brain. So yes, figuring out how to communicate with brain blobs is really useful. I know the title of this paper is saying human level models, but doesn't necessarily mean human level intelligence, although that certainly is the goal for some. We are looking at creating logic for artificial intelligence. It's based off of the way that we think or our brains themselves at the physical level. If you want a wild ride of a paper, look at this one.

[00:01:38]They want to create bioquantum hybrid computers. Yes, they want to combine brain organoids with quantum computers.

[00:01:44]And honestly, I'm here for it. Let's see if it'll work. Now, this was published pretty quietly. You can search that title up there on GitHub. CL1 LLM encoder. This is an independent researcher who rented CL1. So, that is the brain organoid computer that Cortical Labs created. You can purchase it for about 30K. It is a lot. Either way, they used it to integrate it into a language model. This entire workflow could be used for physical reservoirs.

[00:02:09]Our brains are technically physical reservoirs. That is where the physics of the system itself creates logic. This has not been peer- reviewviewed, but I do kind of want to give this all a try.

[00:02:20]I'm interested to see what happens. Now, I do want to tell you that you don't need human brain to be a computer. You can have bacteria be a computer. You can even create bacterial cells for exactly the kind of computation you want. And since logic can arise from their behaviors, that too can be used as a computer. So, we don't really need human brain tissue to be our computers. We can utilize other forms of life. And there are benefits to this. We haven't been able to make true agency, not in silicone, but life encodes randomness.

[00:02:50]Life has desire. So, let's use that. And I know it does seem like cheating a little bit if we're borrowing brain tissue so that it can be artificial intelligence or borrowing bacteria, which arguably have real intelligence.

[00:03:04]Yeah, it's more of a cyborg, isn't it?

[00:03:07]Large language models are not enough.

[00:03:08]They were never enough. We are going to have to embody our artificial intelligence and then they are going to have to make decisions based on the environment that they're in. They're going to have to learn and if they learn, they're going to have to understand the why. An AI needs to understand if it trips over a person that's different than tripping over some laundry. It needs to have morality. It needs to engage with the world and sense it. And it needs to have some skin in the game. Meaning, we're going to have to develop an artificial intelligence system that cares. And us humans do not really know how to do that. And unfortunately we are making robots before we figure out how to do that. So things will go wrong and they are beginning to go wrong. One of the big ideas is that we have to have modular systems. So it has to have multiple layers of cognition in order to become truly intelligent. So it needs to sense the world. It needs to have metacognition. It needs to dream in some circumstances.

[00:04:02]Yeah. Anthropic did make a press release that they are now having Claude dream, which is actually a great idea. And I've seen that integrated into neuromorphic AI. So once it finishes for the day, it's going to process all the information that it learned and just pick out what is important and integrate that. This is thought to be one of the reasons that humans dream. I mean, we may sleep so that we can clear out metabolites from our brains, just junk that builds up in our cells, but we may also use it in order to process the information that we've taken in that we haven't figured out what to do with because we've had so much other stuff to pay attention to. So, yes, a truly intelligent AI will probably need to have lobes not dissimilar to our brains and different processes. I mean, different parts of our own brains take on different tasks. You have your brain stem which is taking on autonomic behaviors. You have to beat your heart and you have to breathe. But you have other parts of your brain that are responsible for cognition.

[00:04:58]So when it comes to LLMs, just give them a break. They may not have sensation yet, but we could give them that. They don't have metacognition, but we could give them that. And eventually, I think us humans will make them so similar to ourselves that we really will have to have the conversation about whether or not an artificial intelligence might have rights. We are not there yet. And I realize there's not a whole lot of nuance in these conversations because, you know, AI tends to promote delusions.

[00:05:24]So, we have to be careful about how we speak about these things. But, I do believe we have a future where something wacky could happen. I hope you've enjoyed this. If the robots ever rebel, I will probably be the first person to tell you about them, and I'm on their side.