AI Summary
5 min readIn a lab in Australia, 200,000 human brain cells fused onto a silicon chip learned to play the video game Doom over the course of a single week. The cells, derived from donated skin and blood, were converted into stem cells, then cultured into neurons on a microarray chip. They were wired directly into the game and taught themselves to play through feedback—when they made the right decision, specific neuropathways lit up, and they learned to repeat that action. The entire setup cost about $35,000, required minimal energy, and the cells picked up the game instinctively, as if the training knowledge was already baked into their DNA.
The efficiency argument against silicon
The hosts argue that this experiment exposes a fundamental inefficiency in how the AI industry currently operates. A standard GPU rack can consume around 700 megawatts of power. The human brain, by contrast, runs on 20 watts. The 200,000 neurons used in the Doom experiment required a fraction of the energy a silicon-based system would need for a comparable task. The cells also learned far faster than a traditional AI model—a week versus the massive training runs and data centers required for large language models. The hosts suggest that the AI industry may have been building intelligence the wrong way: instead of assuming silicon is the best substrate, perhaps the most efficient path is to use organic neuro
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Limitless Podcast
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **The Biological AI Thesis** - Introduces the core idea that organic human/animal cells may be superior to silicon for AI, citing two recent breakthroughs: human brain cells playing Doom and a fruit fly brain uploaded to a laptop.
- 2 (01:28) **Cortical Labs and Dishbrain History** - Traces the company's progression from 2021's Dishbrain (800,000 nerve cells) to 2022's mini-brains playing Pong, to the current Doom-playing breakthrough.
- 3 (02:18) **How Human Brain Cells Learned Doom** - Explains the experimental setup: skin/blood cells converted to stem cells, then to brain cells, cultured on a microarray chip, and wired into Doom for a week of training.
- 4 (03:13) **Three Key Advantages of Biological Computing** - Highlights the major benefits observed: dramatically lower energy use, faster training (one week), and instinctive learning baked into DNA.
- 5 (04:14) **Biological vs. Silicon Learning** - Contrasts how human cells learn (reactionary, feedback-based, no preloaded data) with how LLMs learn (backpropagation, deterministic, silicon-based, inefficient).
- 6 (05:59) **Three Big Implications** - Summarizes the key takeaways: we may be building AI incorrectly, energy needs could be far lower, and biological systems come with a built-in "training run" from evolution.
- 7 (07:28) **Biological Computing as a 10x Accelerator** - Positions biological computing as a novel breakthrough that could rapidly accelerate the AI curve, similar to quantum computing, though still early-stage.
+ Full timestamped outline available in the app
Show Notes
These are groundbreaking advancements at the intersection of AI and biology. Cortical Labs have trained human nerve cells to play video games. Meanwhile, we built a true simulation of a real fruit fly.
We discuss the ethical implications of simulating consciousness and whether these innovations could signal a major shift in AI development.
------
🌌 LIMITLESS HQ ⬇️
NEWSLETTER: https://limitlessft.substack.com/
FOLLOW ON X: https://x.com/LimitlessFT
SPOTIFY: https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQ
APPLE: https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890
RSS FEED: https://limitlessft.substack.com/
------
POLYMARKET | #1 PREDICTION MARKET 🔮
https://bankless.cc/polymarket-podcast
------
TIMESTAMPS
0:00 Custom Brain Cells
6:32 Breakthroughs in Biological Computing
9:16 The Simulated Fly
15:33 The Future of Brain-Computer Interfaces
17:43 The Consciousness Debate
21:32 Challenges in AI Development
------
RESOURCES
Josh: https://x.com/JoshKale
Ejaaz: https://x.com/cryptopunk7213
------
Not financial or tax advice. See our investment disclosures here:
https://www.bankless.com/disclosures
More from this podcast
Limitless Podcast →