Talk about a biocomputer It seems like something from the future, however it is an area that has been developing for years. Now, a Swiss company, FinalSpark, has decided to present a cutting-edge device, a ‘living computer’. We are talking about 16 mini-brains grown in a laboratory from human neural stem cells integrated into an interface.
We have heard thousands of times that despite how advanced computing and artificial intelligence are, the idea of imitating the processing of the human brain to operate a computer is still a challenge. An abysmal leap. So companies like FinalSpark, instead of emulating, have chosen to combine software, hardware and biology to create a pioneering alternative to silicon chip computing. Thus, according to the company, they have achieved a bioprocessor with neurons that can learn and process information.
Functioning
All of this is housed in Neuroplataform, this biocomputer details that it is made up of 16 organoids that are housed in four sets, which are connected to eight electrodes each and a microfluidic system that supplies them. essential nutrients to maintain tissues neurons alive and operational during 100 days.
It would thus be the first online platform in the world that provides access to biological neurons developed in a laboratory, since the company has allowed free access to Neuroplatform to almost a dozen research institutions. The purpose is for them to test these bioprocessors and observe their effectiveness first-hand. While other organizations have already joined the queue of requests that want access to this technology that breaks the mold. “As demand for our neuroplatform grows, we are prepared to meet it, all with the common goal of building the first living processor in the world,” they state in a press release.
Multiple objectives
With this repertoire, the company aspires not only to innovate but also to achieve a unparalleled energy efficiency, creating a much more ambitious processor in its evolution. And according to FinalSpark, this bioprocessor on its Neuroplatform platform can process information with energy consumption up to a million times lower than that of traditional digital processors.
At the same time, its operation for prolonged periods allows researchers to carry out extensive studies remotely. And in the process, they have managed to collect more than 18 terabytes of data. The FinalSpark scientific publication, titled “Open and remotely accessible Neuroplatform for research in wetware computing,” details the results.
Likewise, the development of increasingly advanced computers implies a growing demand for resources, which is why many wonder if they will be sustainable in the long term. His carbon emission and its use of water to keep systems cooled and running is worrying. Hence, the scientists responsible for this biological computer point out that training a single language model like GPT-3 requires approximately 10 gigawatts/hour, around 6,000 times the annual energy consumption of an average European citizen.
While the option proposed by the Swiss company’s researchers would allow the thirst for resources to be curbed. According to ‘Science Alert’, compared to the consumption of GPT-3, the human brain operates 86 billion neurons using only 0.3 kilowatts/hour. And the goal of this company is to get closer to this idea.
This work is preceded by other recent studies that have been showing great potential and are heading in the direction of FinalSpark. This is the case of the project that connected neurons to electrical circuits to create a device capable of recognizing voice. Both are signs of an increasingly firm advance towards an era of computing that until now we only believed could be possible in science fiction. FinalSpark tries to break that prejudice by giving a significant step in that direction.
