tmsr - Search: minsky

mircea_popescu: artificial intelligence as actually extant today is, leaving aside all sorta obscure lulz alf used to like referencing about expert systems and nato wargames, and the assorted lulzpile of neural networks and etc from the very dead 70s, and also leaving aside the whole life OF MINSKY, and then sussma

mircea_popescu: (note that the tempting "obvious" approach -- describe foo then!!! -- is not only fucking broken, but broken in the exact way minsky wasted life trying to produce. there can not be ~description~, the only way to induce meaning in the machine is through filiation. v-produced foo has a very strict "wtf is it ??" answer associated, but also very fine and not structure-driven.)

a111: Logged on 2018-05-14 23:01 mircea_popescu: mit says minksy must use the color green. minsky ignored. mit spent 500 politruk-hours to discern whether he even did what they said or not ; came up with either of these. minsky still ignored.

asciilifeform: *minsky's method

mircea_popescu: that minsky had or had not "his" "bank account" zeroed is ENTIRELY not related to minsky.

mircea_popescu: for random reasons mit decides minsky is no longer allowed in the building ; sussman gave him his key card.

mircea_popescu: mit says minksy must use the color green. minsky ignored. mit spent 500 politruk-hours to discern whether he even did what they said or not ; came up with either of these. minsky still ignored. ☟︎

a111: Logged on 2018-04-12 20:48 asciilifeform: spyked: if you recall, back in the 'minsky age', that was the initial attraction of mechanical 'reasoners' -- discovery of ~simple~ inferences

a111: Logged on 2018-04-12 18:35 asciilifeform: http://btcbase.org/log/2018-04-12#1797532 << as far as i can tell the 'rsa has structure! but aes, surely not' is instance of minsky's empty room ( http://btcbase.org/log/2014-11-13#920444 )

asciilifeform: spyked: if you recall, back in the 'minsky age', that was the initial attraction of mechanical 'reasoners' -- discovery of ~simple~ inferences ☟︎

a111: Logged on 2014-11-13 23:07 mircea_popescu: In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6.

asciilifeform: http://btcbase.org/log/2018-04-12#1797532 << as far as i can tell the 'rsa has structure! but aes, surely not' is instance of minsky's empty room ( http://btcbase.org/log/2014-11-13#920444 ) ☝︎☝︎☟︎

mircea_popescu: i suppose i should have used minsky instead, sussman is a) still extant and b) kinda moved away from ai in the idyllic sense towards more practical applications IE SOLD OUT!

asciilifeform: this was what minsky was trying to achieve with neural network. failed.

phf`: “One day Trurl decided to build a reddit user, but he was afraid that Minsky was going to built one before him. Meanwhile Minsky was building a reddit user of his own, and was trying to hide him from Trurl. etc.

a111: Logged on 2016-05-31 19:46 asciilifeform: this is 'minsky's dark room' all over again.

asciilifeform: this is 'minsky's dark room' all over again. ☟︎

phf: at some point, when i have something to write about. i would've liked to do minsky eulogy for example, but i knew very little about the guy

assbot: Farewell, Marvin Minsky (1927–2016)—Stephen Wolfram Blog ... ( http://bit.ly/1ScjwGN )

ascii_butugychag: mircea_popescu: minsky's proof bulldozed funding for 'ai' for many years; often called the first 'ai winter'

ascii_butugychag: minsky, incidentally, also invented confocal microscope. and ocular display.

ascii_butugychag: btw minsky was quite possibly the most intellectually honest fella in the whole field

assbot: Marvin Minsky - Scientist - 'I want an anti-fad fad' - Web of Stories ... ( http://bit.ly/1RL2i2S )

asciilifeform: 'But, in one of his last interviews, with MIT Technology Review last year, Prof Minsky said there had been "very little growth in artificial intelligence" in the past decade, adding current work had been "mostly attempting to improve systems that aren't very good and haven't improved much in two decades". By contrast, he said, "the 1950s and 1960s were wonderful - something new every week". And he hinted he was against large t

asciilifeform: aha, as pictured in hofstadter and minsky

mircea_popescu: minsky's probably the more advanced work on theoretical neurology to date.

ascii_field: mircea_popescu: are you an aficionado (or the reverse) of m. minsky's 'society of mind' thing ?

phf: i know there's some tension between yaron minsky's (of jane street capital) extensions to core library and inria's core, but someone with better knowledge of language can chime if inria is in the wrong

phf: there's greenblatt, tom knight, david moon, daniel weinreb, guy steele, peter norvig, abelson&sussman, jack holloway. stallman and minsky. not to mention cracauer, eller, fahlman, maclachlan, rme, ron garret of the later people that i know of, tons of people i'm forgetting.

mircea_popescu: not because it wasn't done. but because the idiots doing it (yes, minsky and all) were idiots about the one thing that matters, which had little to do with their actual target.

asciilifeform: undata: the sense of "myself" in my head is a very small chunk << see the crackpot work 'society of mind' by marvin minsky.

mircea_popescu: and yes, one of the many different reasons bayesianism is little more than "a cult of ai without moon or minsky aka stupid"

asciilifeform: !s minsky

mircea_popescu: In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. ☟︎