When researching my book Brainjacking on the science of human use of story to inform, influence and manipulate, I came across an intriguing article in The Times. 'AI will read your mind sooner than you think' blared the headline on Rhys Blakely's piece. He describes the interpretation of signals from electrodes in the brain, processed by AI to help a disabled person be more independent, which is fascinating. But we are then told 'But what about the rest of us? Would you allow a computer direct access to your thoughts? The question isn't as far fetched as it once was.'

Hype or accurate observation?

Blakely describes an experiment at UC Berkeley where subjects were played music - with the help of AI it was possible for listeners to identify the music from processed brain signals, and even pick out some of the words.

I looked into the detail of this experiment for the book. Firstly, given that introduction, you might think this involved detecting your brain activity using some sort of cap, or even remotely. In reality, the experiment relied on electrodes being inserted into the brain. This isn't something I can imagine anyone doing voluntarily unless it was for a medical intervention (and unsurprisingly, here the electrodes were already there for medical reasons - they weren't inserted to do the experiment).

As an aside, when interviewing Mark Gomes, whose novel Age of Extinction featured workers being fitted with chips in their brains, I asked if he really thought it would be come commonplace for people to have surgically implanted chips for non-medical reasons. He replied 'Yes. Not because people want it—but because some will feel they have no choice. In wealthier countries, it’ll be sold as a lifestyle enhancement: faster cognition, seamless access, performance gains. But in poorer parts of the world, it’ll be about survival. If someone offered you a neural chip and, in return, your family could eat for six months—what would you say? That’s not science fiction. That’s economic reality.' 

I'm still not convinced - this is arguably a lot worse than selling a kidney, say, for money.

Back with the real life mind-reading experiment, I was able to access the original paper and from that the audio files produced from brain signals. Here's a challenge for you. This clip is supposed to enable you to recognise the piece of music. And this clip apparently makes  it possible to distinguish some of the lyrics. 

In all honesty, I have not been able to do either, even though I know what's the answers should be. In giving talks on Brainjacking, I've played the clips to my audiences... and no one has ever succeeded in identifying either the song or the words. (If you want to see how you did, scroll down to the bottom of the page for the correct answers.)

I am not suggesting the experimenters were in any way fraudulent, but is it really possible that they have got this right when hundreds of people have failed to come to an answer? I might not be reproducing the experiment under the lab conditions, but I think playing the clips as part of a talk, is a reasonable failed attempt to reproduce the result.

Arguably, the problem, as far as I can tell from the paper (and please don't access it yet if you intend to try the experiment as the song is identified early on), is that the experimenters did not ask even a small number of people, let alone a large sample, to perform the recognition test. Instead their recognition analysis appears to have been based on spectrographic similarities. (Apologies if I missed an actual test with people, but there is far more information on the original sources and processing than on the testing.)

Equally, it's hard not to take the way that the experiment was portrayed in the newspaper as hype. Blakely was just as capable as I was of trying out the actual sound files and seeing if the results made any sense - but there is no suggestion in his piece that there was a problem with this. I will, though, agree firmly with the last sentence of his article. 'And don't let anybody drill a hole in your head.' Certainly for non-medical reasons.

Image by Alex Shuper from Unsplash+

These articles will always be free - but if you'd like to support my online work, consider buying a virtual coffee or taking out a membership:

See all Brian's online articles or subscribe to a weekly email free here What was it all about?
Scroll down...vvvvvvvvFurther...vvvvvvvNearly there...vvvvvvv
The music was Pink Floyd's Another Brick in the Wall, Part 1 - included below for those not familiar with it - and the alleged recovered partial vocals were 'all', 'was' and 'just a brick'.