The questionable ethics of a brain in a chip

Artificial intelligence is the ultimate science fiction boogeyman. The fear of a disembodied, omnipotent entity acquiring full sentience alongside a hatred for humanity is a trope imbued into the pop culture zeitgeist, from the 1967 short story “I Have No Mouth, and I Must Scream” by Harlan Ellison to an April 17 interview with Elon Musk urgently warning of an AI “civilization destruction.” This fear seems like mere fiction, but perhaps the possibility of a tech-fueled catastrophe isn’t so far-fetched.

That’s the question raised by a new scientific venture that looks toward creating super biocomputers using lab-grown brain organoids to replace computer chips. Scientists should not pursue this endeavor until complicated ethical questions are answered. 

While the term “brain organoid” may conjure images of organs locked in jars and hidden away in a mad scientist’s lair, the reality is less grim. Brain organoids don’t resemble tiny versions of the human brain. Instead, they are the size of the tip of a ballpoint pen and composed of stem cells obtained from adult humans via blood and skin samples, according to Dr. Thomas Hartung, a professor of environmental health and engineering at Johns Hopkins University.

The cells are grown in petri dishes to reach an embryo-like state and develop neurons capable of rudimentary cognitive functions like memory and learning. Brain organoids can help identify the mechanisms that underlie neurodevelopmental, neurodegenerative and infectious diseases, like a 2023 study published in Molecular Psychiatry showing that organoids could model the physical effects schizophrenia forges on the brain.

The Catch-22 of this uncanny innovation: The better the proxies resemble human brains, the better they show the effects of diseases, but they also become closer to consciousness — making it harder to justify using them for our selfish purposes. It’s one thing to poke and prod at a clump of cells that experience no sensation. It’s another thing entirely to try to cultivate sentience.

Cortical Labs in Melbourne, Australia, achieved something remarkable in a 2022 study: scientists taught a thumb-size dish of 800,000 brain cells grown on a silicon chip linked to a computer how to play the 1970s arcade game Pong. Researchers hoped to discover a way to program a conscious biocomputer through a brain that learns through experience rather than lines of code.

To prompt the cells to learn the game, the scientists used incentives. If the organoid won, it was met with a nicely organized burst of electrical activity. If it lost, the result was a chaotic stream of white noise. This ability to learn through incentivization is inherently human, with psychologist Burrhus Frederic Skinner identifying it as a fundamental principle of human behavior in the 1940s with his coining of the Incentive Theory. 

In 1981, Hilary Putnam described a famously frightening thought experiment: A “brain in a vat,” snatched from its skull by a crazed scientist who stimulates nerve endings to create the illusion that nothing has changed. How are brain organoids different?

The major difference between AI and a potential biocomputer is that AI’s capabilities are limited to data sets and prewritten code managed by humans. You cannot put a cap on a living organism’s ability to learn.  

“Are there limits to how accurate we want to make these things?” asked bioethicist Insoo Hyun of Case Western Reserve University.

There definitely should be. A 2017 study by biologists at Stanford University revealed a horrifying piece of information about these manufactured creations: They emit brain waves. Researchers found that organoid neurons fired at frequencies indistinguishable from those that animate actual brains, resembling “signatures present in preterm human infants.”   

“In a vat, no one can hear you scream,” writes Stanford law professor Henry Greely in an article published in The American Journal of Bioethics in 2020. How do we know the extent of organoid consciousness? We can’t, and that’s the problem.

In 2021, Oxford philosopher Julian Savulescu and Monash University philosopher Julian Koplin proposed a potential research policy, suggesting that researchers should create no more organoids than necessary, making them only as complex as required to meet the goals of a study and using them only when the expected benefits warrant potential harm.

“If it looks like a human brain and acts like a human brain,” Greely said in his 2020 article. “At what point do we have to treat it like a human brain — or a human being?” 

The time is now. We must start limiting the use of organoids to avoid potentially subjugating a helpless sentient being to experimentation as a means to serve our needs for increased computerization.

This article was submitted as an assignment for the Reporting Captone course at the University of Texas at Austin taught by professor Kevin Robbins.