Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[GiglameshDespair]

More realistic?

Doubtful.  The kind of technology to wire and maintain a human brain, body-free, inside a machine and controlling it doesn't currently exist in any meaningful form, and indeed may never exist due to the ethical complications of, well, chopping people up and sticking them in computers, even if it could be done, and that is something you have to take into account when arguing if something is realistic or not. It's hard to say how much that technology would mass because it simply doesn't exist.

It's also hard to say the precise effects of g-force on a body-less brain because obviously that's never been tested. Generally G-force causes people to pass out when it drives blood away from the brain. Vertical g-force is much worse than horizontal g-force, of which the highest survived amount measured was 214g.

If you want humanoid control, it's going to mean remote control or sending a human - with all their bits attached, most likely.
For something like a probe just send a robot.

[Reelya]

On a side note, how do human brains rate compared to computers, as spaceship cores?  Ignoring the psychological and ethical problems, brains are very efficient computers that last for decades.  How much mass and power would life support and shielding would a brain need per unit of computation compared to a computer?  I know the power efficiency is many orders of magnitude better, but how about mass efficiency?
I guess this leads into a more general point about organic computers vs. silicon.

Human brains don't actually rate all that well for this task. Try being a human and performing all the computation tasks done in a second by a smartphone. Sure, the human brain is more powerful by many measures, but the smartphone will outperform you in most real-world situations all the same. You can also turn a smartphone CPU off to save power and they don't rot. As for "lasts for decades", the US military is still using hardware from the 1970s.

By the time we actually have the tech to put a human brain into a computer, we'll have much better computers than we have now, too.

[wierd]

What the mammalian brain excels at is sensory processing and integration.

A tiny mouse brain does a better job of object identification and spacial orientation than do many high end computer vision systems, which consume much more energy, and take much more space.

But, again, you dont need a human brain for that.

[Madman198237]

I don't know the state-of-the-art, but every bit of material I've ever read suggests that the mouse STILL does a better job than ANY of our best computers.

[smjjames]

Well yeah, of course they do, brains have had eons of evolution to fine tune stuff like that because it's essential to survival.

[Reelya]

"does a better job" is task dependent however so you can't just make a blanket statement like that. Try installing Doom on a mouse, or getting a mouse to do your spreadsheets.

For all our vaunted higher efficiency of overall processing power, it's still the case that we're replacing organic things with computer brains because they're cheaper and more reliable.

[wierd]

Agreed-- hence the opening salvo:  They are better at sensory processing and integration.

They are not better, at basically everything else.

[Madman198237]

I was referring exclusively to wierd's comment on the sensory information processing bit, not in general. In general, the human brain is probably still superior.

After all, the human brain can build a computer, but a computer can't build a human brain (or another computer) without a human to tell it exactly how.

[Reelya]

You don't need a human brain to build a human brain. It's a meaningless statement.

EDIT: And the argument was about whether we should use human brains instead of current computers. We shouldn't, because current computers are better at what we use them to do. Try adding up a bunch of numbers really quickly. Then, try accurately adding 1 billion numbers a second, and tell me how you'd use a squishy meat-brain-machine to emulate that.

EDIT²: Reaction times of machines are far quicker. Neurons can fire up to 200 times per second. Current top-tier consumer-grade CPUs do 1 trillion operations a second. Yes, they don't do the same total throughput as a meat-brain for the same amount of energy, but they just react a million times faster, which might account for why they need more energy per operation.

[Loud Whispers]

You don't need a human brain to build a human brain.

Can a computer make a human brain? 3-d printing human brains is the future

I was referring exclusively to wierd's comment on the sensory information processing bit, not in general. In general, the human brain is probably still superior.

After all, the human brain can build a computer, but a computer can't build a human brain (or another computer) without a human to tell it exactly how.

To be fair though, it is a bit of an odd one. Sorta like how a human brain could calculate the speed, velocity and its expected path of a ball being thrown directly at it, and subsequently coordinate a response to avoid being hit in the face with said ball, all faster than a mechanical computer - but chances are most people would not be able to describe that mathematically in the same way a computer could with any comparable measure of speed. You'd need some way to convert all that biological processing power into a format that is usable with machines innit, a sorta brain interface

[smjjames]

Pretty sure a computer would be able to proccess it faster, it's just that you have to train it to recognize that an object is coming at it and tell it what to do with the data. Even then, the computer can still screw up, as we've seen with attempts with automated cars.

[Reelya]

We also have to take total costs into account: training, reliability, testability and scaling across the lifetime of the project. You can program one machine to deal with a task then test it exhaustively in simulations, then replicate the code/data and you have reliable clones of the trained one. That won't work for meat-brains, each one needs to be hand-trained to do the task all over again.

[smjjames]

(This thread has become an impromptu tech and engineering thread, but since computers are still relevant....)

And if something unexpected happens that it doesn't know how to handle or has no viable options to deal with the problem, the computer will in the best case abort and stop what it's doing, or in the worst case freeze up and crash, or in another worst case, just keep on doing what it's doing.

Yes, you can test a wide variety of things and give it a toolkit (virtual or otherwise) to deal with various situations, but AI isn't at the point where it can figure out what to do on it's own if something happens outside the parameters of it's programming. Edit: Actually, through neural networks or evolutionary learning, it probably could.

[PTTG??]

Human G-tolerance is not a limiting factor in spaceflight. Takeoff of earth's surface is a little rough, but it's already as fast as it can get without running into the air too hard. In space, acceleration slows down dramatically because all our best high-efficiency rockets are low-thrust, high-efficiency models.

Again, robots are cool and all, but they have hard limits, even now, of what they can do. A human and all the monkey tax (which is only a few tons per person) is heavy, but has a much, much higher versatility per pound.

Anyway, here's the first video I've seen from an artist asking to go to space.

It's actually not bad, and I would be happy to see him selected. That said, hooo boy this will be an interesting series of videos and most of them are going to be incredibly self-indulgent.

Also WOW does this sound like the plot of a golden age science fiction novel, except the ship is captured by aliens or flung into the the future or something.

[wierd]

Reaction-wheel reorientation of a craft can happen "Snappy damn fast", which you cannot do with humans aboard.