Chess champion and artificial intelligence researcher Naomi Bashkansky has left OpenAI to join Conduit, a startup developing technology that converts thoughts into text.
Bashkansky, 23, resigned from OpenAI on July 23 and joined Conduit the following day as a founding researcher. “We’re building telepathy: thought-to-text models, trained on non-invasive neural data,” she wrote in a post explaining her decision.
Bashkansky holds the Woman International Master title awarded by the International Chess Federation, one rank below Woman Grandmaster. She won the World School Chess Championship in the girls’ under-13 category and the North American championship for girls under 20. She spent the past year and a half as a researcher at OpenAI, where she worked on alignment and model safety.
“My prediction: In a couple years, the main way we’ll talk with our AIs is with our thoughts,” she wrote. “They will not be a super-smart automated intern or coworker that you desperately try to keep up with. They will be a natural, joyful extension of you.”
Bashkansky envisions users wearing a non-invasive neural band that connects to a computer and translates brain activity into instructions for AI systems. In a scenario she set in 2027, the device decodes a user’s reactions while reviewing code and sends them directly to an AI agent.
“I’m not saying words really loudly in my head while getting coffee,” she wrote. “I just read the plots as I normally do, and make coffee as I normally do. It feels like magic.”
By 2030, she predicts that AI companies will train their systems to connect directly with Conduit’s neural representations, allowing users to communicate thoughts that are difficult to express in words, including mental images.
“My encoded thoughts get sent directly to Codex, rather than having to pass through Conduit’s decoder model first,” she wrote. “That means I easily communicate thoughts that are hard to describe in text, like mental images.”
By 2035, Bashkansky believes AI could feel like “a sixth sense and another limb,” becoming a natural extension of the user rather than a separate tool.
“I like that this is literally a human-in-the-loop vision of the future, where AI directly empowers humans rather than replacing us,” she wrote.
Conduit plans to train its models by comparing non-invasive neural activity with what users are doing at the time, such as the text they write. Bashkansky said the company’s approach depends on collecting far more neural data than academic researchers have previously gathered.
“We must scale up our data collection by orders of magnitude beyond what has ever been done in academia,” she wrote.
She said non-invasive methods offer an advantage because few people are willing to undergo brain surgery, while the hardware for external neural monitoring is improving and becoming less expensive. She added that perfect decoding is not necessary for the technology to be useful, comparing neural signals to an imprecise GPS reading that becomes accurate when combined with a map and route.
Bashkansky said she joined Conduit because of its ambitious vision, its team and the research challenges involved. She estimated that if the company eventually becomes a general platform for reading from and writing to the brain, it could be worth more than $1 trillion.
“I feel good about the vision, because it builds towards a more human future,” she wrote. “And I trust the scaling laws, which have held for many doublings.”
“At OpenAI, you can only work on a narrow slice of The Problem, and even on that narrow slice you’re constrained by the existing architecture,” she added. “At Conduit, it’s all greenfield.”