Brain cells on silicon chips: The rise of ‘biological computers'
In March, Australian startup Cortical Labs unveiled what it says is the first commercial 'biological computer'. Called the CL1, the device integrates lab-grown brain cells – derived from human stem cells – with hard silicon.
The CL1 can be used for drug discovery, disease modelling and research into neuroscience and information systems, Cortical Labs' chief scientific officer, Dr Brett Kagan, told me over a Zoom call.
The CL1 fuses lab-grown brain cells with hard silicon. PHOTO: CORTICAL LABS
With its combination of brain cells and silicon, the CL1 is meant to be able to adapt and learn faster than purely silicon-based artificial intelligence (AI).
Dr Kagan said: 'Cortical Labs was started with the question in mind: 'What if we use the most powerful information processor that we currently know of?' And that, ultimately, is brain cells. Whether it's flies, cats or us, we all can do amazing things – with very little power, very little data – using brain cells.'
Indeed, I was surprised to learn that the brain uses just 20 watts of power, equivalent to what a light bulb would require.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
Sign Up
Sign Up
In contrast, training an AI model, such as the GPT-3 large language model, guzzles as much as 1,300 megawatt hours of electricity – enough to power about 130 homes in the US for a year.
The CL1's launch follows a widely publicised paper published in the journal Neuron in 2022, which gave an account of the Cortical Labs team's efforts to train brain cells in a dish to play the arcade game Pong.
'That paper was a proof of concept to see if you could get brain cells in a dish… to process information and do something in a goal-orientated way – in this case, control a paddle to 'hit' a ball,' said Dr Kagan.
'While we weren't that surprised to see that cells could learn and respond… the speed at which they learnt and responded was surprising. We expected it would take much longer to see some meaningful learning, but in fact, it was within minutes.'
Untangling a paradox
Dr Kagan is excited, not just about the current applications of the CL1, but what breakthroughs it could lead to.
In the near future, applications could include personalised medicine – through which scientists could grow a person's cells in a lab and test drugs on them while measuring the cells' response, enabling the treatment to be tailored to that individual.
Dr Kagan believes the use of biological computers for personalised medicine could be just a few years away with proper investment, given that the technical barriers are not high.
In the long term, a biological computer could be used in fields such as robotics, cybersecurity or even systems with 'generalised intelligence', or in the human-like ability to solve general problems, surpassing current AI systems.
'I think the most exciting applications may be ones that I'm not even going to be able to tell you today,' the chief scientific officer said.
Of course, there are a host of ethical issues that come with such work. For instance, some would be concerned if a biological computer would be able to feel pain.
'No, the system can't feel pain – it doesn't have pain receptors (and) it's not set up to feel pain,' Dr Kagan said, adding that the company has been working with bioethicists on how such technology should move forward.
He noted that scientists have grown brain cells in labs for decades, but previously had no means to test the information-processing ability of these cells. What devices like the CL1 do is introduce new possibilities.
Dr Kagan references a famous quote by the American scientist Emerson Pugh: 'If the brain was so simple that we could understand it, we would be so simple that we couldn't.'
Pointing to Cortical Labs' work, he added: 'What we're building here might be a way to overcome that paradox.'
The rise of biological computers is just one of many weird and wonderful phenomena out there. This column routes signals away from the motherboard of regular news and into peripheral curiosities, whether in finance, economics, science, psychology, or even beyond.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Straits Times
a day ago
- Straits Times
OpenAI prepares to launch GPT-5 in August, The Verge reports
Find out what's new on ST website and app. OpenAI's new ChatGPT model will be positioned as an AI system that incorporates distinct models and can perform different functions as opposed to just a single AI model. SAN FRANCISCO - Artificial intelligence pioneer OpenAI plans to launch its GPT-5 model as early as August, The Verge reported on July 24, citing sources familiar with the plans. The new model, which was expected to launch this summer, will be positioned as an AI system that incorporates distinct models and can perform different functions as opposed to just a single AI model. OpenAI did not immediately respond to a Reuters request for comment. The Microsoft-backed startup's GPT-5 will incorporate its o3 model along with other technologies, chief executive officer Sam Altman had said in February, in a bid to simplify its offerings. The startup ultimately aims to merge the o-series and GPT-series models as it looks to create AI systems that can utilise all available tools and handle a variety of tasks. 'While GPT-5 looks likely to debut in early August, OpenAI's planned release dates often shift to respond to development challenges, server capacity issues, or even rival AI model announcements and leaks,' according to the report. REUTERS
Straits Times
2 days ago
- Straits Times
Meta unveils wristband for controlling computers with hand gestures
Designed by researchers at Meta, this experimental technology reads the electrical signals that pulse through your muscles when you move your fingers. SAN FRANCISCO – The prototype looks like a giant rectangular wristwatch. But it doesn't tell the time: It lets you control a computer from across the room simply by moving your hand. With a gentle turn of the wrist, you can push a cursor across your laptop screen. If you tap your thumb against your forefinger, you can open an app on your desktop computer. And when you write your name in the air, as if you were holding a pencil, the letters will appear on your smartphone. Designed by researchers at Meta, the tech giant that owns Facebook, Instagram and WhatsApp, this experimental technology reads the electrical signals that pulse through your muscles when you move your fingers. These signals, generated by commands sent from your brain, can reveal what you are about to do even before you do it, as the company detailed in a research paper published on July 23 in the journal Nature. With a little practise, you can even move your laptop cursor simply by producing the right thought. 'You don't have to actually move,' Dr Thomas Reardon, the Meta vice-president of research who leads the project, said in an interview. 'You just have to intend the move.' Meta's wristband is part of a sweeping effort to develop technologies that let wearers control their personal devices without touching them. The aim is to provide simpler, quicker and less awkward ways of interacting with everything from laptops to smartphones – and maybe even to develop new digital devices that replace what we all use today. Most of these technologies are years away from widespread use. They typically involve tiny devices surgically implanted in the body, which is a complicated and risky endeavor. These implants are tested solely with disabled people who cannot move their arms and hands, and need new ways of using computers or smartphones. Top stories Swipe. Select. Stay informed. Asia Live: Thailand, Cambodia clash with jets, rockets, artillery in deadly border row Asia 9 Thai civilians killed as Thai and Cambodian militaries clash at disputed border: Reports Singapore Boy, 15, charged after being caught with vapes 5 times; ordered to stay 2 years in S'pore Boys' Home Business MOM probing work injury claim flagged by late Sumo Salad boss Jane Lee: MOS Dinesh Business Marina Bay Sands sees record second quarter with revenue up 37% to $1.77 billion Singapore What's key to a good life? Most Singapore residents choose emotional and mental well-being Singapore Over 2 years' jail for man who worked with wife to cheat her then boyfriend of $220k Asia South Korea police raid offices of BTS' agency Hybe over share probe Neuralink, a Silicon Valley start-up founded by Mr Elon Musk, aims to implant chips under the skull, beside the brain. Synchron, run by an Australian neurologist, hopes to implant devices inside blood vessels in the neck. These efforts, like those of many other start-ups, aim to read brain activity directly – an enormously complex process that must be tailored to the individual. Meta's wristband uses a technique called electromyography, or EMG, to gather electrical signals from muscles in the forearm. These signals are produced by neurons in the spinal cord – called alpha motor neurons – that connect to individual muscle fibres. Because these neurons connect directly to the muscle fibres, the electrical signals are particularly strong – so strong that they can be read from outside the skin. The signal also moves much faster than the muscles. If a device like Meta's wristband can read the signals, it can type much faster than your fingers. 'We can see the electrical signal before you finger even moves,' Dr Reardon said. EMG has long provided a way for amputees to control prosthetic hands. But technologies that use the technique as a computer interface are only beginning to mature. In 2012, three Canadian entrepreneurs founded a company called Thalmic Labs, which built an armband called Myo that sent simple computer commands using hand gestures. With the swipe of a hand, for instance, you could switch to a new slide in a PowerPoint presentation. But the company discontinued the product several years later. Dr Reardon started similar research after founding a company called Ctrl Labs with two other neuroscientists he had met in a PhD programme at Columbia University. In 2019, the start-up was acquired by Meta, where it now operates within a research operation called Reality Labs. Although Dr Reardon and his colleagues have been privately demonstrating their technology for years, they are only now beginning to publicly share their work because it is now mature enough for the marketplace. The key development is the use of AI techniques to analyse the EMG signals. After collecting these signals from 10,000 people who agreed to test the prototype, Dr Reardon used a machine learning system called a neural network – the same breed of AI that drives ChatGPT – to identify common patterns in this data. Now, it can look for these same patterns even when a different person is using the device. 'Out of the box, it can work with a new user it has never seen data for,' said Dr Patrick Kaifosh, director of research science at Reality Labs and one of the neuroscientists that founded Ctrl Labs. According to Dr Reardon, who is also known as the founding father of the Internet Explorer web browser at Microsoft, Meta plans to fold the technology into products over the next few years. In 2024, the company demonstrated how its wristband could be used to control an experimental version of its smart glasses, which can take photos, record videos, play music and verbally describe the world around you. Like technologies from Neuralink and Synchron, Meta's wristband could also provide new options for disabled people. Researchers at Carnegie Mellon are testing the wristband with people who have spinal cord injuries, allowing them to use smartphones and computers even though they do not have full use of their arms or hands. Most people with these types of injuries retain the ability to activate at least some of their muscle fibres. This allows the device to read what their brain is trying to do, even though their bodies can't actually make it happen. 'We can see their intention to type,' said Dr Douglas Weber, a professor of mechanical engineering and neuroscience at Carnegie Mellon. In a similar way, Meta's wristband lets you control a computer with the appropriate thought. Merely thinking about a movement is not enough. But if you intend to make a movement, the wristband can pick up on what you aim to do – even if you do not physically move. 'It feels like the device is reading your mind, but it is not,' Dr Reardon said. 'It is just translating your intention. It sees what you are about to do.' NYTIMES
International Business Times
4 days ago
- International Business Times
What is Baby Grok? Elon Musk Announces AI App for Kid-Safe Content Under xAI
July 21, 2025 21:56 +08 Tech entrepreneur Elon Musk has unveiled plans for a new AI-powered application designed specifically for children. The upcoming app, named Baby Grok , is being developed by Musk's artificial intelligence company xAI. He announced the project in a brief post on X (formerly Twitter), stating: "We're going to make Baby Grok @xAI, an app dedicated to kid-friendly content." X The announcement follows Musk's ongoing criticism of mainstream social media platforms and their effects on young users. In a post last year, Musk warned that "a lot of social media is bad for kids" due to how algorithms compete to stimulate dopamine-driven engagement. He urged parents to monitor and limit children's screen time. Though few details about Baby Grok have been made public, Musk said the app draws its name and creative inspiration from Baby Groot, a beloved character from Marvel's Guardians of the Galaxy franchise. He did not disclose any launch date or feature set but emphasized the goal of promoting safe and positive content experiences for children in a digital world dominated by adult-oriented AI tools. The announcement comes on the heels of xAI's recent release of Grok-4, its most advanced AI chatbot to date. Musk claims Grok-4 can tackle highly complex tasks—ranging from PhD-level academic problems to real-world engineering challenges—with no reliance on existing online data. The model has reportedly outperformed competitors like Google's Gemini 2.5 Pro and OpenAI's GPT-4 in benchmark tests, including the rigorous "Humanity's Last Exam," where Grok-4 achieved a score of 26.9%. A key innovation in Grok-4 is its use of a multi-agent architecture. Instead of relying on a single AI engine, it distributes tasks across several intelligent agents that work together to solve problems. This approach increases accuracy and adaptability, making it more effective in dynamic or high-stakes environments. Whether Baby Grok will inherit any of Grok-4's architecture remains unknown. However, Musk's initiative signals a growing interest in child-safe AI amid rising concerns over online content safety and digital overexposure. As AI continues to evolve, Musk appears keen to position xAI not just as a powerhouse of intelligence but also as a protector of younger audiences in the digital era.
