The News
First, there were prompt engineers, then multimodal model designers. Now, the hottest job at AI companies is a neuroscientist.
The early “neural networks” conceived a century ago are the basis of today’s AI boom. But as LLMs have become common ground for tech companies, so has the need to push boundaries to expand their size and complexity. The private sector’s AI ambitions were enough to convince a career scholar, Aldo Battista, to leave his position at New York University’s Center for Neural Science and join Meta as a research scientist in September.
“Academia is fun because you can explore weird, innovative ideas that may not necessarily have an impact,” Battista said. “But at the end of the day, you want to have an impact on people.”
As a fellow at NYU, Battista studied how the brain decides between subjective options, like what to have for dinner. At Meta, he is working on the neural networks that recommend content to people on social media. The workflow is similar to that of academia, he said, except the feedback is instant. “You can measure if people are liking the algorithm more with this tiny modification you made, or not,” he said.
In addition to the work in algorithms and wearables that companies like Meta could use, neuroscience research closely aligns with two of the biggest areas of study for AI companies right now: energy efficiency and interpretability.
Human brains operate on the equivalent of about 20 watts, performing quadrillions of operations per second. The hardware running AI can match that number of computations, but by using significantly more power. As tech companies multiply their compute capabilities, they’re also searching for ways to make their models more efficient, and further understanding the human brain is one potential method. Neuroscience also offers tried and tested techniques to help understand the reasons models make certain decisions — a research field AI companies are increasingly investing in to make systems more transparent and predictable.
Know More
AI companies, with their seemingly limitless funding and inflated salary offers, are a tempting draw for researchers working in academia, especially as the Trump administration continues to pull federal funding.
There’s both a “push and pull,” away from academia and towards the private sector, said Ray Perrault, computer scientist at nonprofit research institute SRI International.
It’s difficult to know the exact impact of funding cuts on the industry, but in June, neuroscience news publication The Transmitter estimated $323 million in grants from the National Institutes of Health had been cut — a “significant chunk” of the agency’s neuroscience budget, according to a former National Institute of Mental Health director.
Companies like Apple, Google, and Neuralink have sought neuroscientists for years, but the hunt has ramped up as firms look for an edge against their competitors, which may lie beyond traditional engineering hires. “The trend for people leaving academia has always existed, just to a lesser extent,” said Matthew Law, who works on post-training at OpenAI, following a research gig at Stanford’s Institute for Human-Centered AI. “Now, [AI companies] are shifting away from recruiting traditional [computer science] majors and towards the broader research base. There’s both supply, and the pool of AI devs is also getting exhausted.”