Further browsing says the typical brain neuron has an average of 7000 synaptic connections. And synaptic signals are weighted, not purely binary, which can probably be modelled as an 8-bit number. So up to 7000*8=56000 bits per neuron firing, so 56000*.6 billion = 33 trillion bits read per watt.

On the other hand, an M2 has 20 billion transistors, and they mostly don't fire each clock cycle either. Even if they fired once every 100 cycles that would be 200 million transistors firing per clock. If each transistor has 2 inputs, it'd take 17000 transistors firing per cycle to match the bits processed per watt by the brain. I was only counting 64 firing earlier, it's probably much more than that. (The M2 uses more than 1 watt. I didn't quickly find a transistor count for any 1W 1GHz chips so I'm just substituting in the M2 transistor count.)

Browsing the web, I see a brain has about 86 billion neurons, they fire on average once every 6 seconds, and it consumes 23 watts, so 0.6 billion firings per watt. Compare that to an efficient computer chip running at 1GHz on 1 watt of power. Each cycle can fire at least one 64-bit instruction, each firing at least one transistor per bit on average, so at least 64 billion firings per watt. Computers are already at least 100x more efficient than the human brain. If we think better than computers, it's due to the algorithm we thinking with, not hardware.

If a developer suggested those same things that copilot is suggesting, for the same reasons, would that be copyright infringement? Developers are sometimes trained by reading preexisting code, which sounds pretty much the same as copilot. If copilot and developers are equivalent, then you don't need new laws, you just need to apply existing rules.

ChatGPT is nerfed by being unable to do math. It can clearly recognize math, and usually knows what formulas to apply, but does calculations right almost never. If it called out to WolframAlpha when it recognized it was doing math it'd get it right closer to 95% of the time. It can't even multiply correctly. A computer is clearly nerfed if it can recognize the need to multiply but can't multiply correctly.

What's more, if it could do math correctly, it could up the importance of doing math-based sanity checks, because doing math correctly would make those sanity checks useful.

Yay! Simple affordable and large impact. Do it.

Me, just now: what is 8374 * 5297 ?
ChatGPT, just now: 8374 * 5297 = 44364178.

Yay! A constitution (a regression test suite) is my vote on the long-term right way to self-governing AI. It should eventually include ways to evaluate the performance of current rules, add new rules, remove or modify old rules.

I tried Chat GPT and its most obvious limitation to me was it couldn't do math. It knew what formulas to use, it could find the correct inputs, it could give a result and was confident in it, but it'd leave out terms and fail to do conversions and not actually multiply right. If it could reliably do arithmetic and algebra it could run physical sanity checks on things it and the user are saying. It's so easy to fix, I figure it must be broken on purpose.

Pretty sure getting rid of cell phones wouldn't help with aging ... oh wait ... nevermind.

Everyone with a satellite dish they can aim can send a message back, and some will. Dozens in the US alone will even if the US government says not to. Some have bigger and more powerful dishes than others. I recommend continuously streaming Gilligan's Island.

What SHOULD we do? They'll be able to stomp us flat until we've mastered nanotech and space travel and grown accustomed to it for at least thousands of years. Best hope is we aren't contacted by anyone for several thousands of years yet.

The point of a city is to put lots of people close together. A 1D city is less efficient than a 2D or 3D city. I grant you their building is quite tall and fairly wide, for a building. But a 170km*200m base is the same area as a 6km*6km base, or 7km*7km if you leave room for some streets in between. The 7km*7km version has everyone within 7km of everyone else, instead of up to 170km for this thing. The 7km*7km version doesn't have the traffic bottleneck in the middle that the 170km line has, either. The view isn't as good though.

If you're married and you both commute, the best you can do is live in between those workplaces. One's gain is the other's loss. If only one commutes, living close becomes possible.

Machine computing has already performed transformative, disruptive miracles. AI is just more of the same.

The new King is a bit of a nerd.

Fusion would be nice for big cities and industry. For suburbs and rural, solar cells will be more convenient (no need for a power grid or even anything outside your neighborhood), unless fusion gets REALLY small and cheap. Even for big cities, fusion will have to compete with solar cells, which have a big head start.

It'd be hard to do manned interstellar travel without fusion (hard to grow enough food during the voyage between the stars), but that's not going to be a pressing issue for a few centuries yet.

I'm in favor of government DNA databases of everyone. We ought to have public/private key instead of SS numbers and credit cards, and DNA could be used (plus other tests) to reissue a new private key if an old one gets compromised. DNA should automatically be used to confirm paternity. It should be used for medical research. It should be used to identify suspects of crimes, like this. It should be illegal to kill or stalk individuals, but it already is illegal. It should be illegal to harrass people just because they have allele XYZ, but we've already got such laws for racial discrimination, so this also is not new. DNA seems like a different type of photograph. I'm not sure if the DNA database of everyone should be fully public, or only partially public, but the use cases I know for a partially public one seem to me to be all in favor of having it.