I thought so too, until I started my PhD back in 2017. The topic was the application of machine learning to neurosignal decoding. We are absolutely at the top

I thought so too, until I started my PhD back in 2017. The topic was the application of machine learning to neurosignal decoding. We are absolutely at the top of the s curve in this field. There has been no real progress in two decades. The AI boom has just made things worse by redirecting all of the grant money away from better sensors (which could get results in certain niche applications) and towards new algorithms, which have hit an accuracy ceiling because the data their human operators feed into them is complete shit. The field is so beyond saving that an algorithm that achieves slightly better performance on a single old dataset due to sheer chance is presented as a breakthrough. Data rates would be in the tens of bits per minute with error rates of 20% on a good day and no feasible plan to improve either of these woeful specs. I'll be dead before we have a BCI with enough bandwidth that anyone but a dying MND patient would want to get one installed.