61–70 of 115 posts
The melody is all over the place, and the rhythm is hard to tap a foot to, but one thing is certain: it's absolutely convinced that Christmas trees get decorated with flowers. Lots and lots and lots of them.
The melody is all over the place, and the rhythm is hard to tap a foot to, but one thing is certain: it's absolutely convinced that Christmas trees get decorated with flowers. Lots and lots and lots of them.
I could have mistaken it for a track off The Shaggs'[0] lost Christmas album. I like to believe Frank Zappa would have liked this, too. [0] https://en.wikipedia.org/wiki/The_Shaggs
And it just so happens that I live two towns from Fremont, NH, where they're from.
Earlier quoted context omitted.
> the entire source code for the human brain has to be strictly less than 700 MB, because the fully sequenced human genome which obviously encodes the full human mind is less than 700 MB uncompressed Actually those 700MB are compressed, in a way so sophisticated that we don't really know how to uncompress it yet - or whether it's even possible without the external resources our planet provides. And keep in mind that…
What I mean by uncompressed is that you're reading the cytosine (C), guanine (G), adenine (A), or thymine (T) as being two bits per base pair (1 of 4 possibilities). There are 3 billion base pairs, which is 6 billion bits, divided by 8 to get bytes you get 750 MB. The human genome is somewhat redundant and can be further compressed. That is just the string of "ones and zeros" (ACGT) could be run through whatever comp…
Nor can you guess at that encoding by modeling DNA base pairs as 2 bits. DNA base pairs aren't bits. They don't define op codes or memory registers. They are read and executed in a complex way that we don't fully understand. They interact with each other in complex ways that we don't fully understand.
There's way more information encoded in those base pairs than a simple 2 bits. You can't simply model them that way and declare that there are X Bytes in a genome. DNA isn't digital.
The melody is all over the place, and the rhythm is hard to tap a foot to, but one thing is certain: it's absolutely convinced that Christmas trees get decorated with flowers. Lots and lots and lots of them.
Earlier quoted context omitted.
Instead of five, don't you mean "two to three"? And under that comparison, isn't it scary how much it does sound that way? Like a kid who hears words together but don't know what they mean yet? (and doesn't really get the cultural things around it.) To me it sounds like a quite musical 2-3 year old stringing words together. Doesn't it strike anyone else that way? These things are going to grow up very, very soon! We…
> uses some 20 watts Daniel Lemire has an observation here: our brains don't use all the connections, all the time (when it does it's called epilepsy); it would be interesting if we built some computers that also worked like that - with only small parts being active at a certain time, but switching quickly from one to another. That would help with the heating problem and it might allow processors a lot more complex t…
Earlier quoted context omitted.
"Musical sense" is a pretty subjective thing, which I suspect is one of the primary issues with artificial creativity. That said, assuming training data comes from music with progressions that could be broadly classified as 'popular music,' you would expect to find some regression to the mean with more production and deeper training data. (to abuse a phrase) One other issue that I think will come up is how insular an…
The statistical approach is painfully naive and doesn't work - as is obvious from the example. It's like feeding a net with the complete works of Shakespeare and expecting it to produce a genius-level original play. It's simply not going to happen.
"The best Christmas present in the world is a blessing." This algorithm is throwing down some wisdom
It kind of feels like a five-year-old trying to make a song. Which seems good - now they only need to improve the mechanisms and who know, maybe we'll get to the point when it's a ten-year-old?
Instead of five, don't you mean "two to three"? And under that comparison, isn't it scary how much it does sound that way? Like a kid who hears words together but don't know what they mean yet? (and doesn't really get the cultural things around it.) To me it sounds like a quite musical 2-3 year old stringing words together. Doesn't it strike anyone else that way? These things are going to grow up very, very soon! We…
In a sense, that's expected, and it's a (perhaps weak) indication that we're moving in the right direction.
> You're watching a two-year old. It's interesting to think about what it will be like when it's ten, sure. But what will really blow your mind is what it'll be like when it's 23.
Or 23000. Because, you know, it's a machine.
> the entire source code for the human brain has to be strictly less than 700 MB, because the fully sequenced human genome which obviously encodes the full human mind is less than 700 MB uncompressed
I have this half-baked argument sloshing around in my mind for a while now, that the total source code is greater than that, because it also includes constants, rules, and facts built into the fabric of the Universe itself. In other words, that 700 MB is code for "build a carbon-based intelligence in the context of such-and-such universe" - where "such-and-such" may turn out to be a surprisingly long and intricate description, effectively extending the code by a large amount. Kind of like a fixed ROM added to the variable NVRAM of all humans.
In a different universe the same code will produce a very different creature - and that's at best. In most other universes, the process would terminate early with a critical error, or would not even start. Or would not even make sense at all.
> we are at an incredible pivot point in human history
I think we are at the cusp of an explosive change without precedent in the history of the Earth - and not just due to AI. There's a long list of things that are about to hit tipping points in the next years / few decades. It will be... "interesting". Buckle up.
Earlier quoted context omitted.
What I mean by uncompressed is that you're reading the cytosine (C), guanine (G), adenine (A), or thymine (T) as being two bits per base pair (1 of 4 possibilities). There are 3 billion base pairs, which is 6 billion bits, divided by 8 to get bytes you get 750 MB. The human genome is somewhat redundant and can be further compressed. That is just the string of "ones and zeros" (ACGT) could be run through whatever comp…
I think that your "700MB uncompressed" fails to take into account that the construction, development, and maturation of the brain relies heavily on cellular and molecular mechanisms. I think it is a little disingenuous to hide the enormous wealth of information necessary to create a brain, much less understand and utilize one, inside of your compiler.
Earlier quoted context omitted.
How long did it take to go from really silly and obviously impractical flying machine designs to working, practical ones?
Thousands of years? Who knows how long before the Wright Brothers people have been attempting flight?