Written by Julian Savulescu, Chris Gyngell, Tsutomu Sawai
posted with The conversation
Julien Savulescu, University of Oxford; Christopher Gygel, The University of Melbourneand Tsutomu Sawai, Hiroshima University
It’s 2030 and we’re at the biggest tech conference in the world, CES in Las Vegas. A crowd gathers to watch a major tech company unveil its new smartphone. The CEO arrives on stage and announces the Nyooro, containing the most powerful processor ever seen in a phone. The Nyooro can perform an astonishing quintillion operations per second, which is a thousand times faster than smartphone models in 2020. It is also ten times more energy efficient with a battery that lasts ten days.
A journalist wonders: “What technological advancement has allowed such enormous performance gains?” The CEO replies: “We have created a new biochip using human neurons grown in the laboratory. These biochips are better than silicon chips because they can change their internal structure, adapt to the user’s usage pattern, and lead to huge efficiency gains.
Another reporter asks, “Aren’t there ethical concerns about computers that use human brain matter?”
Although the name and storyline are fictional, it’s a question we have to deal with now. In December 2021, Melbourne-based Cortical Labs developed clusters of neurons (brain cells) that were incorporated into a computer chip. The resulting hybrid chip works because brains and neurons share a common language: electricity.
In silicon computers, electrical signals travel along metal wires that connect different components together. In the brain, neurons communicate with each other using electrical signals through synapses (junctions between nerve cells). In Cortical Labs’ Dishbrain system, neurons are grown on silicon chips. These neurons act as the wires of the system, connecting different components. The main advantage of this approach is that neurons can change shape, grow, replicate or die in response to system demands.
Dishbrain could learn to play the arcade game Pong faster than conventional AI systems. Dishbrain developers said, “Nothing like this has ever existed before… It’s a whole new way of being. A fusion of silicon and neuron.
Cortical Labs believes its hybrid chips could hold the key to the kinds of complex reasoning that today’s computers and AI can’t produce. Another start-up making computers from lab-grown neurons, Koniku, believes its technology will revolutionize several sectors, including agriculture, healthcare, military technology and airport security. Other types of organic computers are also in the early stages of development.
While silicon computers have transformed society, they are still surpassed by the brains of most animals. For example, a cat’s brain contains 1,000 times more data storage than an average iPad and can use that information a million times faster. The human brain, with its trillions of neural connections, is capable of performing 15 quintillion operations per second.
This can only be matched today by massive supercomputers using large amounts of power. The human brain only uses about 20 watts of energy, about the same amount of energy as it takes to power a light bulb. It would take 34 coal-fired power plants generating 500 megawatts per hour to store the same amount of data contained in a human brain in modern data storage centers.
Companies don’t need brain tissue samples from donors, but can simply grow the neurons they need in the lab from ordinary skin cells using stem cell technologies. Scientists can engineer cells from blood samples or skin biopsies into a type of stem cell that can then grow into any type of cell in the human body.
However, this raises questions about donor consent. Do people who provide tissue samples for technology research and development know that they could be used to make neural computers? Does he need to know this for his consent to be valid?
People will undoubtedly be much more willing to donate skin cells for research than their brain tissue. One of the barriers to brain donation is that the brain is seen as tied to your identity. But in a world where we can grow mini-brains from virtually any type of cell, does it make sense to draw this kind of distinction?
If neural computers become mainstream, we will face other tissue donation issues. In Cortical Lab’s research with Dishbrain, they found that human neurons learned faster than mouse neurons. Could there also be differences in performance depending on the neurons used? Could Apple and Google be able to make super-fast computers using the neurons of our best and brightest today? Would anyone be able to obtain tissues from deceased geniuses like Albert Einstein to make specialized limited-edition neural computers?
Such questions are highly speculative but touch on broader themes of exploitation and compensation. Consider the scandal involving Henrietta Lacks, an African-American woman whose cells were widely used in medical and commercial research without her knowledge and consent.
Henrietta’s cells are still being used in applications that generate huge revenue for pharmaceutical companies (including recently to develop COVID vaccines. The Lacks family has yet to receive compensation. If a donor’s neurons end up being used in products like the imaginary Nyooro, should they be entitled to a share of the profits from these products?
Another key ethical consideration for neural computers is whether they could develop some form of consciousness and feel pain. Would neural computers be more likely to have experiences than silicon-based ones? In the Pong experiment, Dishbrain is exposed to loud, unpredictable stimuli when it gets a wrong answer (the racket misses the ball) and to predictable stimuli when it gets it right. It is at least possible that a system like this can begin to experience unpredictable stimuli as pain and predictable stimuli as pleasure.
Scientific Director Brett Kagan of Cortical Labs said:
The fully informed consent of the donor is of paramount importance. Any donor should be given the opportunity to reach a compensation agreement as part of this process and their bodily autonomy should be respected without coercion.”
As discussed recently in a study, there is no evidence that neurons in a dish have qualitative or conscious experience, so they cannot be distressed and without pain receptors they cannot feel pain. Neurons have evolved to process information of all kinds – being left completely unstimulated, as is currently done all over the world in laboratories, is not a natural state for a neuron. All of this work is just allowing neurons to behave as nature intended at their most basic level.
Humans have used animals to perform physical labor for thousands of years, although this has often led to negative experiences for the animals. Would using organic computers for cognitive work be more ethically problematic than using an ox to pull a cart?
We are in the early stages of neural computing and have time to think about these questions. We need to do this before products like “Nyooro” go from sci-fi to mainstream stores.
Julian Savulescu, Visiting Professor in Biomedical Ethics, Murdoch Children’s Research Institute; Emeritus Visiting Professor of Law, University of Melbourne; Uehiro Chair in Practical Ethics, University of Oxford; Christopher Gyngell, Biomedical Ethics Researcher, The University of Melbourneand Tsutomu Sawai, Associate Professor, Humanities and Social Sciences, Hiroshima University
This article is republished from The Conversation under a Creative Commons license. Read the original article.