Advancements in brain-computer interfaces (BCIs) are accelerating rapidly, raising both promising possibilities and ethical concerns about the future of human-AI interaction. A recent 168-page essay titled “Phase Lock” by entrepreneur and investor Saturnin Pugnet has sparked significant discussion within AI and neurotechnology circles by exploring how BCIs could potentially enhance human cognition and facilitate closer integration with artificial intelligence.
BCIs, which are already employed medically to treat conditions like Parkinson’s disease and enable paralyzed patients to operate computers, might one day offer what Pugnet describes as "cognitive superpowers." He argues that in an era of rapidly advancing AI, slowing down neurotechnology development may not be feasible or desirable. Instead, speeding up human cognitive capabilities through BCIs could be essential for humans to maintain relevance alongside increasingly intelligent AI systems.
This perspective aligns with earlier ambitions such as Elon Musk’s Neuralink project, launched in 2016, which aims to develop implantable BCIs to vastly increase the rate of information exchange between human brains and computers. Musk has highlighted the immense disparity between computer processing speeds—measured at trillions of bits per second—and human brain processing, estimated at just a few bits per second. Bridging this gap, he suggests, would require merging biological and digital intelligence.
Investment in neurotechnology has surged accordingly, with venture funding more than doubling from $2.3 billion in 2024 to $4.8 billion last year. Pugnet proposes that neurotechnology could benefit from the same scaling principles that powered the development of large language models (LLMs) in AI. These models process massive datasets to recognize patterns; similarly, AI could analyze extensive brain activity data to better understand and even augment cognitive functions.
However, Pugnet acknowledges significant risks. He warns that without ethical safeguards, the technology could be misused or cause unintended harm. Nonetheless, he suggests that integrating brain data into AI training could lead to safer models that better reflect human values through a process he calls “alignment by integration,” potentially more effective than current “alignment by control” methods.
Practical examples of implanted BCI technology include a recent study published in Nature Medicine, which documented a paralysed 45-year-old man composing over 180,000 sentences at an average speed of about 56 words per minute using a BCI device for speech and cursor control. Meanwhile, non-invasive brain-imaging technologies are also progressing, such as the British start-up Cerca Magnetics, which recently secured £7 million in government funding to deploy brain-scanning services at two NHS hospitals to aid epilepsy treatment.
Despite these advances, some experts remain cautious. They point out the complexity and variability of brain signals, which can differ widely depending on a person’s emotional or physical state, making the reliable decoding of abstract thoughts a formidable challenge. The idea that AI could not only interpret but also transmit information directly to the brain remains speculative.
Nonetheless, venture capitalists continue to back these ambitious projects, accepting that even if they fall short of their most transformative goals, they will generate valuable data sets that could significantly advance medical research.
As neurotechnology and AI converge, the debate continues over how to balance innovation with ethical oversight, and whether the potential cognitive enhancements will ultimately serve human interests or introduce new risks.