It's not just Elon building bridges to the brain: The Internet of Things is coming to a head Technical work demands tools. Software developers have integrated development environments and text editors. Genetic researchers have gene sequencing machines and CRISPR. Doctors have too many toys to name. Bryan Johnson, founder and CEO of Kernel, wants to build tools for interacting with the human brain. For all the fawning …
Reading brain functions doesn't bother me, but writing? Not to go all tinfoil hat here, but there's not an open system that can't be broken into with sufficient time, effort and incentive. And when you apply that to the human brain, you literally get mind control. That's a pretty big incentive.
(Paris, because at least she'd be safe.)
"Reading brain functions doesn't bother me"
Tell that to the US immigration officer who is demanding your social media accounts, your bank accounts, and access to your brain's workings while he interrogates you about your beliefs.
I wonder how nuanced it can be?
Yes there is progress: cochlear implants don't give normal hearing but we are so amazing that they allow the recipient to understand speech, gadgets to control limbs by thought work too, but again because we are amazingly adaptive and clever and can learn to use them.
A brain on life support controlling a waldo-avatar machine styled like an android is still science fiction, though no doubt feasible (unlike a computer AI controlled avatar). See Anne McCaffery Ship books, OTH, Ian M.Banks sort of AI and transhumanism (reading an identity and resurrecting it) seems like egotism by people that can't accept death, not even real SF, but magic fanasty dressed as SF, cute names though.
Like every other tech today, this will be oversold. Hopefully though eventually people that are quadraplegic or in a "locked in" condition will benefit. It will be clumsy and take weeks to years to learn use. Not a clip on and instantly browse the internet, write a novel, video edit or VR gadget, though given time and no alternative I'm sure those are possible.
Isaac Asimov wrote a short story about an autistic lad in Mars orbit who controls a robot on the surface of Mars.
The Good Doctor would also play with sci-fi - in another story a scientist is commissioned to research depression... He ends up, after analysing brain waves, proscribing a blast of 'When the Saints Come Marching In'
"The brain is the most powerful form of generalized intelligence in the known universe"
Perhaps that should read:
"The brain is the most powerful form of generalized intelligence known in the universe"
It is not even a "generalised intelligence".
It is highly specialised towards certain tasks needed for the survival of humans in the Serengeti and their ancestors, such as visual processing and some types of pattern matching. There are plenty of very simple tasks that it is rubbish at, e.g. give me a list of a hundred random numbers, find all the spelling mistakes in the posts in this forum, and memorise and recite the complete works of Shakespeare. Even quite simple computers can do those things, but people can't.
It's pretty darn general. The brain evolved for survival, but this gave us a brain that can learn lots of different skills at the same time. You can learn algebra, French, astophysics, poetry, origami ...
It can also do pretty amazing things without you needing to actually, ahem, *think* about it. A good example is juggling. I can juggle no problem, but if I was to try to work out the maths of juggling I would go mad.
I wouldn't call successfully training muscle memory to "twitch like this, wait approximately that long, catch returning object here, repeat" (basically by random trial and error until you get it right, which is why so many of us never manage to get to three objects) a particularly "intelligent" and/or amazingly high brain function.
@DropBear - There's more to juggling than muscle memory, although it plays an important part. It's hand-eye coordination that matters; just try juggling with your eyes closed. With very limited information about the trajectory of the balls, the brain can extrapolate the trajectories and move your hand to roughly where it needs to be. I think muscle memory is more inportant for throwing the ball on approximately the right trajectory in the first place.
I agree that our brains are just a bunch of specialised tools, but we can combine them to do things that we haven't actually evolved to do. We may not be fully general -- we can't learn everything. But we have a level of generality way beyond any current AI, and general AI is a very hot research area right now.
You can learn algebra, French, astophysics, poetry, origami ...
That is true, but they are all much of a sameness. They are all governed by more-or-less fixed rules, and they are all human inventions. You might argue about astrophysics and algebra, but they are rule-based models that humans have invented to help them to, respectively, describe things and manipulate numbers. People have much more trouble where the rules aren't clear cut, or vary, or are very complex, although people have specialised abilities relating to language, which makes poetry and French easier.
As a concrete example, we might be able to make accurate predictions in astrophysics, but doing the same in macroeconomics seems to be beyond us.
I actually met a guy who was a "slack wire" walker who had a mathematician ask him to teach him how to do it.*
The mathematician did indeed write numerous equations to explain what was happening but the walker said he'd be better off if he just got up on the wire (it's about 50cm off the ground) and got used to falling a few times. That multi billion element multi layer neural network we carry around is quite versatile.
*I doubt that sentence caused most of you any problems comprehending it but it will probably generate 10s of parses in a human language parser, possibly crashing the parser in the process.
Despite the first attempt to apply more or less conventional conventional compiler technology to the problem (IE no backtracking but 3 symbol lookahead, where a "symbol" can be quite complex) was made by Mitch Marcus in the late 1970's.
As for direct neural interface we still can't do eyes, although it seems someone has started to actually look at what the optic nerve produces and work backward to how it produces that signal. Big surprise. It's nothing like a TV signal, digital or analogue.
give me a list of a hundred random numbers, find all the spelling mistakes in the posts in this forum, and memorise and recite the complete works of Shakespeare. Even quite simple computers can do those things
None of those involve intelligence and can in theory be done with racks of relays or even cogwheels.
Who wrote the programs that allows a computer to do any of those? Without a human written program it's just a bunch of expensive sand.
Actually Random numbers are REALLY hard to program. The trivial simple method is a physical noise source such as a pair of neons or zener diodes. A pair so you can cancel out repetitive supply voltage noise.
Or a big predefined table.
A lot of computer based "random number" generators can be predicted if you store enough past output and the generator isn't influenced / seeded by external semi-random events.
Certainly a human can't do random numbers with an equal probability, but possibly can give you random digits that are more random in reality than a "straight" computer program.
Several computational methods for random-number generation exist. Many fall short of the goal of true randomness, although they may meet, with varying success, some of the statistical tests for randomness intended to measure how unpredictable their results are (that is, to what degree their patterns are discernible).
https://en.wikipedia.org/wiki/Random_number_generation
It's really hard without having the program read external events (noise).
Random doesn't even mean what many think it means.
There ARE "write" interfaces, for vision and hearing. The sensation is nothing like regular vision or hearing and the user can take over a year to "learn" the relationship between the real world and what they experience.
So indeed not like video tape, hard drives or flash memory.
Our memories are not remotely like a PVR or RAM.
Using Musk in the title/body come across as trolling. Musk is involved in AI to open source it - his is fearful of what it can do to the human existence, as in ending it. Musk thinks the way to counter this is to give the power to the masses rather than just those that want to monetize/control. If we survive this, we may have him to thank.
Every time you cross a road with traffic coming from both directions, you solve about three second-order differential equations with every millisecond that you're in transit (assuming you don't want to meet the traffic en route). Having programmed computers to solve differential equations, I believe the brain does it faster - because it does it using analogue, not digital techniques. Perhaps some of these researchers should give that a thought...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
