Hacker News Reader

Comment by rockemsockem | Hacker News Reader

In the face of a future where AGI is real it seems to me today the best way to keep humans able to move at the same pace as the AGI is brain computer interfaces like Neuralink. We can have a slightly less than AGI AI that we can communicate with instantly just by thinking. You can work through problems with it, offload number crunching, etc. Seems like what we need. If you just don't like Elon, then you can hope with me that this spurs on other companies to do the same so we can get some options!

View on HN

Sure, and in the face of a world where yogurt cured cancer, it would be irresponsible not to produce as much Chobani as possible. AGI is a sci-fi story molded into an investment pitch.

I would highly doubt that brain computer interfaces will be sophisticated enough to compete with an AGI. The brain is messy and sending large amounts of information from chip to the brain in a way that can be interpreted seems impossible without something really invasive. And if the technology is monetized like most tech, your implant will cost you a monthly subscription service.

The end goal of Neuralink is so short sighted. When Musk proposed it, I first began to question his genius.

At its best, Neuralink is supposed to give humans a chance against AI by speeding up human i/o. But if AI continues to expand its capabilities and speed, then that "advantage" will only last for a short time.

In exchange, we will give corporations and governments direct read/write access to our brains.

Oh, and if AI does take over the world, Neuralink style implants will likely come in very handy to control humans, won't they?

This isn't just about Musk. Ideally, nobody should ever do this. This is the final frontier of agency and privacy.

How is anyone on-board with this?

The better IO will let humans directly interface with non-AGI AI to compete. Imagine if GPT-4 could be accessed just by thinking. Now imagine the same except with GPT-7 or w/e we get to by then. So it doesn't last for a short time, it's a permanent plug in to AI itself.

Corporate and governmental access are extremely scary. Companies will, as always, have to be the first movers in this space, but once it takes off we NEED open hardware and open source software. Imagine running an obfuscated binary on your brain. No thank you.

Do you really see technology like this *not* coming to fruition though? Do you think we will never have mass-market BCIs?

> The better IO will let humans directly interface with non-AGI AI to compete

My point was that there is no competition, even with neuro interfaces. Unless it stops advancing for some reason, AGI will win. If faster I/O helps today, the advantage will be gone tomorrow. If humans have to "compete" with AGI, then we are guaranteed to lose in the long run.

> Imagine running an obfuscated binary on your brain. No thank you.

You say no thank you, but most people won't care if it makes their lives 1% easier. Musk, for example, is historically against open sourcing any of his products.

> Do you really see technology like this not coming to fruition though? Do you think we will never have mass-market BCIs?

This needs to be the first time that we actually regulate an area of tech before it becomes a multi-billion dollar industry. We screwed up on web tracking. That cannot happen again. We need a neuro bill of rights ASAP.

How long until TSA requires a quick scan before boarding?

This is the final frontier of agency and privacy.

Please read the transcript, or listen to this podcast with Sean Carroll and Nina Farahany. There are primitive brain interfaces entering the market now, including one from Meta.

https://www.preposterousuniverse.com/podcast/2023/03/13/229-...

We don't know how AGI will shake out or, especially, what it will even look like. What does out-competing AGI even mean? One of the most positive outcomes I can think of is AGI being aligned with human interests broadly, but it is doing so much that no one knows wtf is going on. A BCI can bridge that understanding gap to at least keep humans in the know and capable of steering the ship. If the competition is AGI wants to kill us, then we're probably hosed anyway.

That's why I said we need open source and open hardware alternatives. But there's no world where that kind of tech comes first. Companies are going to develop the cutting edge first, when has that not been the case?

I'm really, really happy to see that there's healthy competition in the space. Especially, ironically, from Meta who has just been so great on the open-source front for quite a while.

The only reason that megalomaniac Zuck is open sourcing his LLM is because he is behind Open AI, so AI safety be damned.

Please, please listen to the podcast that I mentioned earlier.

If you are down, I have email notifications turned on here. If you listen to that podcast, and the still don't want to regulate the hell out of brain interfaces, I would love to discuss, even weeks later.

Skipping the keyboards and eyeballs might give you a factor of 10, assuming our brains aren't pipelining anything.

Transistors outpace organic synapses by significantly more than humans on foot outpace continental drift, which is a gap so much larger that a mere factor of ten is irrelevant.

Right, that's why you offload significant tasks to transistors, way more than just basic addition, subtraction, etc. Present AI makes it possible to offload more complex tasks to AI altogether with minor attention paid for checking and revising, etc.

That analogy ultimately goes as well as continental drift offloading movement onto humans, and the AI doing to humans whatever the analogy is of humans doing this:

https://en.wikipedia.org/wiki/Mountaintop_removal_mining

https://en.wikipedia.org/wiki/Bucket-wheel_excavator

What makes you think he will succeed?

I’m still waiting for any sort of progress towards the hyperloop. Not to mention "Full" self driving, or Mars colonies. Last time I heard of Neuralink, they butchered hundreds of animals for no apparent progress.

I can understand cynicism in general, but the Mars stuff — all of SpaceX really — is going so much faster than the never-delivered visions from governments and corporations worldwide for as long as I've been alive.

That (and that the cars shipped instead of being vapourware like all the other "green" initiatives from the incumbents) are why people were even inclined to bother listening to any of the other projects he's been involved with, even though the same ideas would normally be dismissed as if the proponent couldn't tell fact from fiction.

Yeah or the choice is not to hook your brain up or suffer you know?

Found Elon Musk's throwaway account.

Reply on news.ycombinator.com