[infurl]

https://www.lesswrong.com/posts/xwBuoE9p8GE7RAuhd/brain-efficiency-much-more-than-you-wanted-to-know

This article attempts to calculate how efficient human brains might be compared to some absolute physical limits on computation. As the author points out, this is an important question because if human brains are already as efficient as it is possible to be then any artificial intelligence that we invent is going to be similarly constrained. In other words, the smartest machine that we could make is going to be just like a human brain and it's not going to be significantly smarter.

After a great deal of research and calculation the author concludes that the human brain is already within roughly an order of magnitude of optimal.

Achieving those levels of energy efficiency will probably require brain-like neuromorphic-ish hardware, circuits, and learned software via training/education. The future of AGI is to become more like the brain, not less.

[MagnusWootton]

Yeh I agree,  the robot wont be much different.    It will be more efficient at all tasks,  do everything in super ways,  but its just a superfluity without much real important gain.  The main parts of the difference of intelligence we already have.

[Zero]

If this is about energy efficiency, it doesn't mean we can't do better in terms of pure performance. For example we can build Formula One cars. They are not "energy efficient" at all, compared to, say, a cheetah (for the same same speed the F1 consumes a lot more energy), but it is wayyyyyyyyyy faster. Same goes for spatial efficiency, ...etc.

Am I wrong?

[MagnusWootton]

Computers dont take much power to run.   The computer design itself can have alot to do with the energy expenditure,  even if it has 10,000,000 parallel lines.  But you cant really talk about it with the amount of stupid bullies on the internet that already know what they are doing.

I'm saying it here cause aidreams is a little different than say  eevblog,  which is full of dummies that already know everything about what has already been done. (and its the same on every electronics engineering site on the net, they are ALL DUMB!!! all of them!)  But its a guaranteed fact none of them designed a cpu before.  except...  they do make fpga designs...  this doesnt make sense...

At the moment I'm scared to mention on a thread that an ac square power supply is not more complicated than a sine wave one,  but I daren't say because of the ego's I need to protect there.    You just convert to dc then oscillate the signal.  its just a tiny bit added to the end of an ordinary generator.

[MikeB]

Synapse speed varies greatly depending on the situation... You can be thinking critically outside, then walk into a room with 60hz flashing bulbs and forget what you were doing... Sometimes you can hold a higher synapse speed than lights and see them flickering... So using a 0.5hz synapse speed is bizarre.

The writer of the article doesn't compare the minds efficiency for any task, such as navigation in a hot desert/sauna. Only theoretical process speed...

The last thing is (about AGI being knowledge based) that when you were young and didn't know anything about the world, were you unable to have an opinion on anything (eg fairness, direction) or were you a repeating machine...