Gone awfully quiet!

None of the above. It looks for examples of the same code having been written before, and mashes up something to present to you. It doesn't 'know' about speed or portability or code size, absent somebody remarking about those in its input data.

There's a lot of code out there, so if you ask for a fast algorithm then it can probably dredge one up, but it doesn't 'know' why it's fast.

If you ask it why it's fast, it can look for somebody talking about that in the training data and present that as an argument, but it doesn't guarantee to relate to the same code example it provided.

For toy problems that have been done a million times before, the training data is pretty solid so it might look good. But once you start going off track into areas the training data is sparse then you can look more closely.

Theo

And electronic design is not just coding. It needs real, organic intelligence.

It's impressive that a human brain only needs about a hundred watts.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

To do it well. More or less adequate electronic design is easier. I've cleaned up after few people whose idea of "adequate" fell a bit short.

It is woefully slow.

That must be the 234,412,265th time you've said that.

At some things. Not at others.

That may be something of an exaggeration.

Name one. The basic operations in the human brain seem to work in the millisecond range, and ECL can do stuff in a nanosecond.

The human brain does well on large data sets because it has a lot more parallel processing than a regular computer, but - as demonstrated by the solution of the protein folding problem - you can get more data into a big computer than you can into a human brain.

Ok 234,412,104th

Understanding a language you're fluent in appears to be near enough instant. Why would you need it any faster? I've never seen ECL do that.

Sure but I've yet to see an online AI which learns from its mistakes. And when that happens, who is going to teach it what a mistake is and what isn't? Some subjects, such as politics, may run into the same difficulties humans have. Where will a DT made of ECL with a much larger data set lead us?

Some very impressive things can happen in milliseconds.

Sometimes complex things are processed in background, and can take days or even years.

What's cool is that one can have a problem, forget about it for years, see some new component that makes it work, and have a new circuit pop up instantly.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

John Larkin is easily impressed by his own brilliance,

If you have a very slow brain.

That just lots of memory. One time I did that was when an impractical way of dealling with ripple on pulse width modulated output which I came up with in 1975 became practical in 1992 when I got my hands on a big-enough chunk of programable logic - not all that big, as it was a plug-in replacement for a 22V10 chip, but big enough. Obviously, I hadn't forgotten about it. I hadn't been obssessing about it for the previous 17 years, but I hadn't forgotten about it either.

I am impressed by my brain, the one I was born with.

Or if you allow it to work at all time scales.

So, you have a very slow brain?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I don't think our brains are a lot different from the ones our ancestors had 5,000, or 50,000 years ago. So why did evolution make them/us able to do calculus and design electronics and program in Rust?

It's assumed that, since brains are such energy hogs, critters don't evolve much more brain than they really need. And most don't.

Humans benefit from making fire and making weapons, but those wouldn't need the ability to do abstract math.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Most egomaniacs are. It's fairly prominent feature of the condition. Trump has described himself as a "stable genius" which is a comic illustration of the egomaniac capacity for self-delusion.

You do have some conscious control of what your conscious mind does. The sub-conscious is less accessible.

The human brain doesn't seem to any kind of delay line store. Stuff gets encoded, and you can decode it when you need it. I do find myself remembering stuff from sixty or seventy years ago, so there may be some kind of house-keeping processing sorting through the memory banks in background.

I have met Elizabeth Loftus, and know that this gets complicated when there's significant emotional content, but I'm not getting a lot of that.

I take no credit for having my brain, and I'm impressed by most anybody's brain.

DT does seem to have created a lot of peace and saved a lot of lives, so far.

She sounds awful.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

More if they go to they trouble of flattering you.

It does get him the attention he craves. The Gaza riviera was something of a false start - but the images of Trump resort hotels along the Mediterrainean coast must have been seductive.

She was great company. My wife had known and liked her for years.

Chomsky thinks that our capacity to use language to communicate depends on fairly recent tweaks to our brains. Human language is a more complicated communication system than anything else we've looked at, and presumably this lets us move to a higher level of abstraction than our competitors. When we got to be able to talk about mathematics we'd got into a more productive region than any other creature we know.

But if there's an ecological niche that a big brain can exploit, brains will get bigger.

They got a lot more from cooperative hunting and defense. Dunbar's number is 150 which means that we live in bigger packs than most social mammals. Language lets us coordinate even bigger groups.

Some people don't like that, and Trump does seem freeze out experts who don't know him well enough to be aware of his need for flattery.

Well you don't seem to be able to separate anything from personality so why should AI?

Ask the hostgages.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

You were starting to have a sensible discussion.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I've been wondering what arguments DT might have used to achieve this. It's not his charming personality, for sure.

Jeroen Belleman

More like a gross exaggeration.

I wonder what you think you means by that? And any intelligence I have is entirely natural, so my antics aren't any kind of guide to what artificial intelligence might do. Intelligence is about drawing conclusions from data - personality is more about the kinds of conclusions you want to be able to draw, which famously biases the sort of data you will go to the trouble of collecting. The easiest way of seeing it in action is to let different personalities look at notionally identical data sets, and compare their conclusions.