AI Models Built From Rat Brains Just Got Closer to Reality

以下正文同步自 Wired，版权归原站所有，已转换为易读排版。

Starting Tuesday, select Amazon Web Services customers will gain access to The Biological Computing Company’s “rat brain” AI model as part of a limited preview. The startup’s technology is specifically designed to improve AI for generating videos. Both Amazon and The Biological Computing Company, which goes by the acronym TBC, says they expect the tech to roll out to all AWS enterprise customers soon.

“We figured out a way to code information, like images for example, to the biological material,” says TBC cofounder Alexander Ksendzovsky. “We then observe how the biology processes that information, and then we build a tool that mimics that process.”

This is not the first biologically-derived computing platform that Amazon has made available in its marketplace, says Deap Ubhi, global director of technology for startups at Amazon Web Services. Ubhi says the cloud-computing giant also works with Cortical Labs, an Australia-based company that combines lab-grown neurons with silicon chips to help companies process data. (It calls its products WAAS, or “wetware as a service.”) Cortical Labs also sells a multi-thousand dollar “biological computer,” a low-power device designed for use in laboratories that can supposedly keep neurons alive for six months.

TBC takes a “pragmatic approach,” says Ubhi, which is part of why the company appealed to Amazon. Rather than taking big swings or trying to reinvent the transformer, the core architectural unit of large language models, “they’re working within existing standards of the generative AI space and saying, ‘How can we make the current visual models more efficient?’” he explains.

As AI companies race to find ways to make their models more efficient, a once-fringe field known as biological computing has begun gaining traction. Researchers have long envisioned a world where computer software performs more like the neural networks inside human brains instead of relying solely on math-based algorithms.

But biological computing comes with unique challenges. The field requires running actual biology labs, not just computer labs, in which brain cells, stem cells, or synthetic biomaterials must be kept alive or preserved, carefully monitored, and somehow translated into meaningful digital information. While some startups have made progress in the field, bridging the divide between nature and code remains complicated.

The Biological Computing Company was founded in Baltimore, Maryland four years ago by two neuroscientists and neurosurgeons: Ksendzovsky, now the company’s CEO, and Jon Pomeraniec, who serves as president and COO. Earlier this year, TBC raised $25 million from a group of investors led by Primary Venture Partners, its first significant round of funding.

Shortly after that round closed in March, the company secured an additional $25 million, bringing its funding total to more than $50 million. This new round of capital has not been previously reported.

Last year the startup opened offices and a small research and development lab in San Francisco, where 35 employees work directly with rat brain cells as well as human stem cells. The living cells are placed on multi-electrode silicon arrays made by 3Brain, a Swiss biotech company. Researchers use the arrays to send electrical stimulation patterns into the neurons and record how they respond. TBC then analyzes that neural activity for computational patterns that can be translated into software and used to improve existing AI models, specifically video generation models.

The decision to focus on video out of the gate was both scientific and practical, says Ksendzovsky. One factor was the physical layout of the multi-electrode silicon arrays Biological Computing Company uses. The place where each electrode sits affects how it interacts with the neurons. The company realized images were a natural place to start, since visual information could be mapped onto the grid more readily than text or language.

From there, a potential business use case emerged. The startup theorized that by modeling how the neurons process images, the insights could be used to improve AI models for generating video.

Pomeraniec says that Jeff Dean, a prominent AI researcher and TBC investor, was the person who first suggested that the company try fine-tuning video generation models before trying to apply its neural tech more broadly.

Because there’s already accepted industry benchmarks for video models, the startup could show early on whether its tech represented a meaningful scientific advancement by testing its performance on problems that have been solved already. “Jeff said if we could do that, there’s the promise of doing more complex, interesting things with our AI models down the line,” Pomeraniec explains.

Previously, TBC’s technology was only available through a neocloud provider called Bluesky Compute. The startup is claiming that its AI model, when compared against the open-source AI model it runs on, is up to five times faster at video generation and significantly lowers the cost of inference, the processing or “thinking” part of an AI model versus the training of it. (The company won’t say which open-source AI it is using as a comparison, but claims it’s one of the “frontier” video generation models.)

Now that Amazon is distributing TBC’s software, the ambitious startup could potentially reach a much larger customer base. Amazon’s Ubhi says he’s excited about The Biological Computing Company, but it remains to be seen how well its technology can scale.

“I think with any model like this, the one thing you have to ask is, if you push it to the extremes, will you still see improvements? If you have customers who want to generate longer-form videos, ten minutes, an hour, what’s the potential loss in fidelity over the course of time?” he asks. In other words, a model needs to be able to keep track of what has already happened and maintain consistency as a video gets longer. As TBC enters its next phase, the company will be looking to see whether its technology holds up as new customers try pushing it to the limit.

正文由 FLUX 从来源站点 RSS 同步，内容未经改写；遇到排版缺失或需要图片、视频时请以原文为准。