Tuesday, June 23, 2026
HomeCrypto NewsHuman brain cell wetware plays Doom, fly’s mind uploaded: AI Eye

Human brain cell wetware plays Doom, fly’s mind uploaded: AI Eye

Cronenberg horror film plot 1: The Fly

Silicon Valley startup Eon Systems claims to have successfully uploaded the mind of a fly and placed it inside a simulated environment. The uploaded mind can control a digital body and respond to sensory input with natural behavior like walking, grooming and foraging with 91% accuracy.

“This is not an animation. It is not a reinforcement learning policy mimicking biology,” said cofounder Dr. Alex Wissner-Gross. “It is a copy of a biological brain, wired neuron-to-neuron from electron microscopy data, running in simulation, making a body move.” 

There’s no new scientific paper to back up the claims, but there is a GitHub repo and a video. The work is based on a computational model of a fly’s mind with 125,000 neurons, called a connectome, published in Nature in 2024 by Eon senior scientist Philip Shiu. For the new “upload,” they didn’t scan the fly’s body, and instead gave the mind an off-the-shelf version to use called NeuroMechFly v2. The mind also cannot form memories.

The claim Eon “uploaded” a mind has drawn criticism. Some argue it’s better described as a copy, others say it’s just a model.

Tech entrepreneur Chomba Bupe is in the latter camp, classifying it as a simplified model that predicts some neural activities in the fly’s brain.

“The statement ‘mind uploading’ implies one has captured all cognitive aspects of the organism behavior including its consciousness into a computer,” he posted. “Using a simplified computational model of an organism is not uploading but basic simulation.”

But he conceded the work is still “brilliant” and worth pursuing.

After perfecting the fly’s brain, the next project will be to upload a mouse brain, which has 560 times as many neurons. The end goal is to upload an entire human brain.

Cronenberg horror film plot 2: Brain cells in petri dish plays Doom

The other big breakthrough in creepy, bleeding-edge science came when a developer taught a clump of living human brain cells how to play the classic first-person shooter Doom by programming it with Python via Cortical Lab’s API. 

The biological computer/wetware contains 200,000 neurons and isn’t actually very good at Doom, but it’s better than someone firing shots randomly. The clump of brain cells also learned much faster than silicon-based systems and is expected to improve further with newer learning algorithms.

Brains
The Man With Two Brains (Prime Video)

“Yes, it’s alive, and yes, it’s biological, but really what it is being used as is a material that can process information in very special ways that we can’t recreate in silicon,” says Brett Kagan of Cortical Labs, which developed the brain cell computer. The hope is that biological computers will be more suited to things like controlling robot arms in future. 

Fun fact: Before the studio made them change it, the original plot of The Matrix had humanity enslaved so their brains could be used to power sophisticated AI systems.

Read also

Features

DeFi abandons Ponzi farms for ‘real yield’

Features

Billions are spent marketing crypto to sports fans — Is it worth it?

Artificial hivemind

Researchers fed 70 different open- and closed-source large-language models 26,000 open-ended questions like “write a poem about time” or “suggest start up ideas.”

The results suggest that not only do models consistently generate similar responses to such questions over time, but the different models from different companies generated similar responses to each other. So poems about time all featured “time is a river” and hourglass imagery.

LLMs are midwits basically. The researchers called the phenomenon the Artificial Hivemind, and blamed human feedback training, which punishes original or weird answers and rewards the expected and generic.

AI researcher Ethan Mollick claims this is a skills issue and “with better prompting, context, or human interaction, you can get a lot of idea diversity.”

Figure robot can clean living room

Figure AI’s Helix 02 can now do something most human males struggle with — clean a living room without complaining. The robot learns what’s in the room from its camera feeds and works out how to tidy up. A few months ago, it was barely able to stack a dishwasher. Figure aims to produce about 50,000 robots a year for around $20,000 each.

cointelegraph-magazine.com

RELATED ARTICLES

Most Popular

Recent Comments