Wherein a Twitter user dispels the illusions of machine learning and artificial intelligence.
Here’s the news report on the mystery. It turns out a YouTuber illegally uploaded clips from a fifteen-year-old Japanese TV show that featured “Over Your Shoulder” by Dinosaur Jr. Japanese viewers started to binge on the videos, and social media created a self-reinforcing feedback loop. When Dinosaur Jr. charted above Ariana Grande, the YouTuber got spooked and scrubbed all the clips from his channel. That left Billboard scratching their heads as to how the anomaly happened.
Pretty fucking terrifying, yes.
“Machine learning” is rhetorical at best and more literally a contradiction in terms. “Learning” is predicated on an intellect, and man can never build a machine with a true rational intellect. Intelligence isn’t a matter of sufficient complexity or processing power, as shown by the potentially disastrous results of entrusting scientific research to massive lumps of code.
As should always have been apparent, the scientific method’s efficacy relies on scientists’ informed intellects, and its accuracy relies on their freedom to run valid experiments. Since both the intellect and the will are non-material faculties, it doesn’t matter how much silicon you stack up. Only trained human beings can do science.
A corollary to the thread above pointed out by another Twitter user: Science is as self-correcting as accounting.
None of this is to say that we won’t have increasingly sophisticated Chinese room style “A.I.”, and it’s always fun to speculate about artificial intelligence in science fiction. Soft A.I. figures prominently in my martial thriller series Combat Frame XSeed. Read it now:
"As should always have been apparent, the scientific method's efficacy relies on scientists' informed intellects, and its accuracy relies on their freedom to run valid experiments. Since both the intellect and the will are non-material faculties, it doesn't matter how much silicon you stack up. Only trained human beings can do science."
It also counts on their commitment to serve the data, serve reality, rather than succumb to the temptation to be the high priests of naturalism.
Note: I use the following functional definition of religion. Any worldview or belief system that provides answers to past, present, and future. 'Where do we come from?' 'What should we be doing now?' 'What will happen to us?'
It appears to be a flavor of gnosticism, but the secret knowledge is of the natural, not the supernatural. I suspect the logic runs something like this:
1. Science deals with the natural, so everything science can explain is natural.
2. Science can explain everything.
3. Therefore, everything is natural.
4. So all claims of supernatural are false.
5. I know this to be true and others don't, so it is my duty to share truth with them…
Scientist as priest goes back at least as far as Foundation.
William M. Briggs calls it scidolatry.
That is wonderful. Dialectic and rhetoric combined into a truthful meme.
It goes back far in Fandom. I'm covering a series on it now:
https://wastelandandsky.blogspot.com/2019/02/fandom-illustrative-history-part-i.html
You're doing yeoman's work, sir!
Thank you!
I read the article, and I had subconsciously come to a similar conclusion: science fiction is about setting, rather than plot, and I have been assigning genre based on plot more than setting.
It was a very interesting read!
Related: my meditations on the GPT-2 text generator, considered too dangerous to release by its creators.
Harry Potter and the Portrait of what Looked Like a Large Pile of Ash
It's the most interesting the music industry has been in years, and it had to be a meme centered on song from a time when music made them money.
This is truly the winter phase of Western civilization–a sort of funeral elegy for the brighter past.
One of the advantages of being old enough to have written Fortran programs on Hollerith cards is that this is not my first A.I. rodeo. Computers are very good at pretending to be intelligent and humans seem to have an inborn ability to be fooled by them. Whether it's a simple syntax processor like Eliza, or a search engine with a natural language interface like Watson, we tend to see them as thinking beings.
They are no such thing of course, and can never be. You can program a computer to play chess or be a contestant on Jeapordy, but you can't program it to understand what it is really doing. The computer will only really "know" the mathematics of the rules of the game, not why the game is being played.
We humans do not (and will never) understand the whole consciousness thing. Without that understanding, hard A.I. is a myth. Anyone telling you different is selling you a bridge.
Interesting you bring up Eliza.I just watched Moon the other night, and in the trivia they pointed out that Gerty's dialogue was based on Eliza.
Yep, only a hundred lines of code or so (most of it canned responses) and it had folks telling it some of their most private thoughts. Amazing.
Chris Lopez:
"They are no such thing of course, and can never be. You can program a computer to play chess or be a contestant on Jeapordy, but you can't program it to understand what it is really doing. The computer will only really "know" the mathematics of the rules of the game, not why the game is being played."
Roger Penrose's "The Emperor's New Mind" did a pretty thorough job of debunking Hard AI, though he couldn't bring himself to follow though on what those premises implied.
Chris Lopez:
"We humans do not (and will never) understand the whole consciousness thing. Without that understanding, hard A.I. is a myth. Anyone telling you different is selling you a bridge."
The amusing thing about Hard AI proponents is that they sold themselves that bridge first. They may be the remnants of the druids, come back to worship at the standing stones.
They are very close to the leaders of Lewis' organization N.I.C.E., worshiping naked at the feet of the Head, asking it for its wisdom.
"Only another line of code, then Utopia!"
Brian,
So Collapsing Empire is prophetic with regards to relying on artificial intelligence?
Moral of the story people still matter and computers are just tools and AI isn't a form of nu consciousness
xavier
Remember the old saying "to err is human, but to really screw up you need a computer". To those of us who worked with them back then, the truth of it made it even funnier. Computers will do EXACTLY what you tell them to do, whether it makes sense or not.