{"id":32784,"date":"2025-06-29T04:25:42","date_gmt":"2025-06-29T04:25:42","guid":{"rendered":"https:\/\/financialrush.com\/?p=32784"},"modified":"2025-06-29T04:25:42","modified_gmt":"2025-06-29T04:25:42","slug":"inside-the-british-lab-growing-a-biological-computer","status":"publish","type":"post","link":"https:\/\/financialrush.com\/?p=32784","title":{"rendered":"Inside the British lab growing a biological computer"},"content":{"rendered":"

\n<\/p>\n

\n

In a laboratory outside Cambridge sits a remarkable \u201cbiological computer\u201d. Its 200,000 human brain cells, grown in the lab, lie on silicon circuitry that communicates their synchronised electrical activity on a screen to the outside world.<\/p>\n

The CL1 device, about the size of two shoe boxes, was developed by Australian start-up Cortical Labs with the UK\u2019s bit.bio, in a bid to create \u201csynthetic biological intelligence\u201d \u2014 a new form of computing that could offer opportunities beyond conventional electronics and other developing technologies such as quantum.<\/p>\n

\u201cLike our brains, biological computers will consume many orders of magnitude less energy than conventional electronics as they process information. Future applications could include robotics, security and the metaverse,\u201d Cortical Labs chief executive Hon Weng Chong told the Financial Times. <\/p>\n

The fast-growing search for alternatives to energy-intensive conventional electronics has stimulated the new field of biological computing<\/a>, which aims to tap directly into the intelligence of brain cells rather than simulating it in silicon through \u201cneuromorphic\u201d processing and AI. <\/p>\n

Cortical Labs is at the forefront of this movement, though academic groups and other start-ups such as Swiss group FinalSpark and Biological Black Box in the US are also making progress. <\/p>\n

<\/picture>
Inside the CL1 biological computer, human neurons grow on top of silicon circuitry<\/span> \u00a9 Chris Radburn\/FT<\/span><\/figcaption><\/figure>\n

Early applications of CL1 are in neuroscience and pharmaceutical research, discovering how different chemicals and drug candidates affect the brain cells\u2019 information processing. <\/p>\n

\u201cThe next stages of innovation will make possible new and more advanced forms of computation beyond conventional AI systems, using the same processors \u2014 neurons \u2014 that underpin intelligence in living organisms,\u201d added Chong. <\/p>\n

For Mark Kotter, clinical neuroscience professor at Cambridge university and bit.bio founder, the significance of CL1 \u201cis that it is the first machine that can reliably assess the compute power of brain cells. That is a real paradigm shift.\u201d<\/p>\n

Experts noted that CL1 was a \u201cremarkable achievement\u201d, that has helped advance the fledgling biological computing field. <\/p>\n

Karl Friston, a neuroscience professor at University College London who has also collaborated academically with a number of Cortical Labs scientists, said it could be regarded as the first commercially available biomimetic computer. <\/p>\n

\u201cHowever, the real gift of this technology is not to computer science \u2014 at the moment. Rather, it is an enabling technology that allows scientists to perform experiments on a little brain.\u201d<\/p>\n

Professor Thomas Hartung of Johns Hopkins University in Baltimore, who is investigating \u201corganoid intelligence\u201d<\/a> using cerebral organoids or mini-brains grown from stem cells, said the outstanding contribution of Cortical Labs was to develop virtual games-playing as a benchmark for biological computing. <\/p>\n

CL1\u2019s predecessor, called DishBrain, learned to play the simple video game Pong, in which it moved a virtual paddle up and down to deflect a ball. <\/p>\n

Training involved giving the neurons a \u201creward\u201d stimulus when they moved the paddle correctly, by applying electrical activity in the form of a sine wave, which the cells like. The \u201cpunishment\u201d when they got it wrong was unpleasant white noise.<\/p>\n

\"Close<\/picture>
Microscope view of neurons growing on electronic circuits<\/span> \u00a9 Chris Radburn\/FT<\/span><\/figcaption><\/figure>\n

Experiments with DishBrain and CL1 show how different conditions affect the neurons\u2019 information processing, measured by how well they play Pong. \u201cWe have treated them with chemicals that have an impact on our brains,\u201d said bit.bio\u2019s Kotter. \u201cThis machine shows for example that alcohol degrades your ability to compute.\u201d<\/p>\n

Another experiment compared the effect of three epilepsy treatments and found that one of them, carbamazepine, was superior in improving gameplay metrics. <\/p>\n

\u201cWe are thinking a lot about how to program our biological computers,\u201d said Chong. \u201cOne big question is how we represent digital information to these neurons.\u201d The scientists are teaching the neurons the shapes of digits, he added, \u201cand they are now starting to recognise that a nine is different from a four or a five.\u201d<\/p>\n

Cortical Labs and bit.bio lay down pure layers of two specific types of neuron on the silicon circuitry of the CL1 biocomputer \u2014 one to excite electrical activity and the other to damp it down. \u201cThe balance between acceleration and brakes is really important,\u201d Chong said. The neurons are grown from stem cells derived originally from the human skin. <\/p>\n

Others such as Switzerland\u2019s FinalSpark are exploring biological computing with cerebral organoids. But bit.bio and Cortical Labs believe their layers of standardised neurons will give more reproducible results than organoids. <\/p>\n

\"Tony<\/picture>
bit.bio scientist Tony Oosterveen collects live neurons from an incubator.<\/span> \u00a9 Chris Radburn\/FT<\/span><\/figcaption><\/figure>\n

\u201cOur neurons look very homogeneous,\u201d said Tony Oosterveen who leads bit.bio\u2019s brain cells work. \u201cIf you look at other technologies you will see huge variation. Our strength is to make pure populations.\u201d<\/p>\n

Whatever the long-term promise of biocomputing, its advocates concede that adoption for more general applications and AI lies decades in the future. One problem is to work out an efficient programming system. <\/p>\n