‘Mind-reading’ AI: Japan study sparks ethical debate

AL Jazeera  :
“I went into the bathroom and looked at myself in the mirror and saw my face, and thought, ‘Okay, that’s normal. Maybe I’m not going crazy'”.
Takagi and his team used Stable Diffusion (SD), a deep learning AI model developed in Germany in 2022, to analyse the brain scans of test subjects shown up to 10,000 images while inside an MRI machine.
After Takagi and his research partner Shinji Nishimoto built a simple model to “translate” brain activity into a readable format, Stable Diffusion was able to generate high-fidelity images that bore an uncanny resemblance to the originals.
The AI could do this despite not being shown the pictures in advance or trained in any way to manufacture the results.
“We really didn’t expect this kind of result,” Takagi said.
Takagi stressed that the breakthrough does not, at this point, represent mind-reading – the AI can only produce images a person has viewed.
“This is not mind-reading,” Takagi said. “Unfortunately there are many misunderstandings with our research.”
But the development has nonetheless raised concerns about how such technology could be used in the future amid a broader debate about the risks posed by AI generally.
In an open letter last month, tech leaders including Tesla founder Elon Musk and Apple co-founder Steve Wozniak called for a pause on the development of AI due to “profound risks to society and humanity.”
Despite his excitement, Takagi acknowledges that fears around mind-reading technology are not without merit, given the possibility of misuse by those with malicious intent or without consent.
“For us, privacy issues are the most important thing. If a government or institution can read people’s minds, it’s a very sensitive issue,” Takagi said. “There needs to be high-level discussions to make sure this can’t happen.”
Takagi and Nishimoto’s research generated much buzz in the tech community, which has been electrified by breakneck advancements in AI, including the release of ChatGPT, which produces human-like speech in response to a user’s prompts.
Their paper detailing the findings ranks in the top 1 percent for engagement among the more than 23 million research outputs tracked to date, according to Altmetric, a data company.
The study has also been accepted to the Conference on Computer Vision and Pattern Recognition (CVPR), set for June 2023, a common route for legitimising significant breakthroughs in neuroscience.