The main goal of science is to obtain new knowledge. And our question here is, is AI able to deliver new knowledge? We don't need theories and hypotheses, we need big data and some algorithms and AI will come up with interesting new knowledge.
New knowledge is closely related to discovering necessary relations between events in the real world. It is all about causality. How do we know that two events that go together, are not coincidental, even though it happens hundreds of times, but have a causal relation?
We already saw in the previous lecture that a strong method to decide on this issue is the use of counterfactuals. Is AI capable of such an epistemological exercise?
When I took the statistics course at the university, the central theme was correlation. You take two sets of events and calculate statistically how often they go together.
If this going together achieved a certain value, there was a correlation, which meant that when you see event A in the real world, there is a high chance that event B will occur too, based on the outcome of your correlation calculation. That was new knowledge about the world.
But was that a causal relation? Not really. Judea Pearl (1936), an Israeli-American computer scientist and philosopher, best known for championing the probabilistic approach to artificial intelligence, wrote "The Book of Why" (2018).
The book explores the subject of causality and causal inference from statistical and philosophical points of view for a general audience. In it, he develops the "ladder of causation".
The first, simplest level is finding associations, nothing more than David Hume's custom. This level is just about looking at events or data that occur together, coincidentally or otherwise.
Like the rooster crowing and the sunrise. Or the size of a house and its price. Or the level of education and consumption of chocolate in a given country.
This is purely about patterns, not about the exact relationship between them. Large amounts of data are perfect for this. A webshop can clearly see that people who buy football boots often also order shin guards, without there being a cause-and-effect relationship between them.
The second level on the ladder of causality goes one step further. This is where interventions come into play. Suppose someone has a headache and then takes an aspirin.
An intervention is then performed and after a while, we look at what happened next. The crucial difference with the first level is that there is only observation, but there is now an active intervention or change.
For example, a retailer no longer just looks at the sales figures but changes the price to see what the sales figures do next. Interventions seem to be a powerful tool, not least because they evoke associations with scientific experiments.
For example, one group receives a medicine, the other group does not, and by comparing the groups one can determine whether the medicine works.
I'll save the third level for the next lecture. Causality is a mysterious phenomenon. It makes us think of a necessary relation between A and B. But what makes it necessary?
To digress a little...what I love the most is the POST HOC PROPTER HOC fallacy. Translating the Latin for you: After this, because of this.
Stupid example: people hear a bell ring and a rock falls from the sky...just coincidental, but from that moment on when people hear that bell ring they run away... post hoc propter hoc ...the bell causes rocks to fall from the sky, they believe.
Main Sources:
MacMillan The Encyclopedia of Philosophy, 2nd edition
TABLE OF CONTENT -----------------------------------------------------------------
1 - 100 Philosophers 9 May 2009 Start of
2 - 25+ Women Philosophers 10 May 2009 this blog
3 - 25 Adventures in Thinking 10 May 2009
4 - Modern Theories of Ethics 29 Oct 2009
5 - The Ideal State 24 Febr 2010 / 234
6 - The Mystery of the Brain 3 Sept 2010 / 266
7 - The Utopia of the Free Market 16 Febr 2012 / 383
8. - The Aftermath of Neo-liberalism 5 Sept 2012 / 413
9. - The Art Not to Be an Egoist 6 Nov 2012 / 426
10 - Non-Western Philosophy 29 May 2013 / 477
11 - Why Science is Right 2 Sept 2014 / 534
12 - A Philosopher looks at Atheism 1 Jan 2015 / 557
13 - EVIL, a philosophical investigation 17 Apr 2015 / 580
14 - Existentialism and Free Will 2 Sept 2015 / 586
15 - Spinoza 2 Sept 2016 / 615
16 - The Meaning of Life 13 Febr 2017 / 637
17 - In Search of my Self 6 Sept 2017 / 670
18 - The 20th Century Revisited 3 Apr 2018 / 706
19 - The Pessimist 11 Jan 2020 / 819
20 - The Optimist 9 Febr 2020 / 824
21 - Awakening from a Neoliberal Dream 8 Oct 2020 / 872
22 - A World Full of Patterns 1 Apr 2021 / 912
23 - The Concept of Freedom 8 Jan 2022 / 965
24 - Materialism 7 Sept 2022 / 1011
25 - Historical Materialism 5 Oct 2023 / 1088
26 - The Bonobo and the Atheist 9 Jan 2024 / 1102
27 - Artificial Intelligence 9 Feb 2024 / 1108
The Discussion
To be continued...thank you for your attention again..
19] Max Chatnoir: It's helpful if you can link some mechanism in the first condition to the following even.
[13:19] Max Chatnoir: event
[13:19] herman Bergson: Yhe What if the approach is more telling...the counterfacual approach
[13:20] Particle Physicist Bejiita (bejiita.imako): aha
[13:20] Max Chatnoir: Yes, if B doesn't follow A, then what?
[13:20] herman Bergson: What if the rooster doesnt crow....>
[13:20] herman Bergson: ?
[13:20] herman Bergson: The sun comes up anyway
[13:20] Particle Physicist Bejiita (bejiita.imako): then everyone starts to scream ITS THE END OF THE WORLD!
[13:20] Particle Physicist Bejiita (bejiita.imako): lol
[13:20] herman Bergson: oh yes...I forgot Bejiita :-)
[13:21] Neuro Wonder: no, everyone uses other variables to ascertain the situation
[13:21] Particle Physicist Bejiita (bejiita.imako): haha
[13:21] Neuro Wonder: and, it's also the end of the world :)
[13:22] Particle Physicist Bejiita (bejiita.imako): in contrast, in programming like in my game dev worlds i make in Unreal Engine and Unity everything is casual but thats because i need to explicit tell what should happen, so there i can make a bell that WILL cause rocks to fall
[13:22] Particle Physicist Bejiita (bejiita.imako): but its only in the computer
[13:22] herman Bergson: But the main issue here is, can AI recognize causality?
[13:22] Max Chatnoir: Watching the eclipse at totality yesterday, I could see how it could be pretty startling.
[13:22] herman Bergson: We can....but AI?
[13:22] Neuro Wonder: yes
[13:22] Particle Physicist Bejiita (bejiita.imako): ah
[13:23] Max Chatnoir: could AI propose a connection from cause to effect?
[13:23] herman Bergson: So far AI has reached the first level Pearl describes....showing correlations
[13:23] Particle Physicist Bejiita (bejiita.imako): hm
[13:24] Neuro Wonder: AI could propose multiple causes, and where experimentation is needed, would tell humans what research needs to be carried out, including step by step instructions. "put chick embryo in this media plate, look through electron microscope at this resolution, etc"
[13:24] Neuro Wonder: When robots can do those things, then we can just tell AI to go ahead and carry out the experiments that robots would perform for us
[13:24] Max Chatnoir: Yes, it might be able to suggest an experiment.
[13:25] Neuro Wonder: Experiments will still have to be done, no doubt
[13:25] Particle Physicist Bejiita (bejiita.imako): ah
[13:25] herman Bergson: Yes it can indeed
[13:26] herman Bergson: It can create molecules in a more efficient way than the lab does sometimes
[13:26] Neuro Wonder: for sure
[13:26] Particle Physicist Bejiita (bejiita.imako): as ong it have the correct patterns that makes it work
[13:26] Particle Physicist Bejiita (bejiita.imako): long3
[13:26] Particle Physicist Bejiita (bejiita.imako): and those need to be casual
[13:27] Neuro Wonder: especially in structural biology to know the exact molecular makeup and orientation of a protein
[13:27] Particle Physicist Bejiita (bejiita.imako): like if u to this then that WILL happen 100 %
[13:27] Neuro Wonder: There is no 100%
[13:27] Particle Physicist Bejiita (bejiita.imako): well almost at least
[13:27] Particle Physicist Bejiita (bejiita.imako): there need to be a proven connection
[13:28] Neuro Wonder: yes, there will always be error bars in science
[13:28] Neuro Wonder: 99.99999% confidence is enough for me
[13:28] Neuro Wonder: especially if i'm just building a deck in my backyard lol
[13:28] Particle Physicist Bejiita (bejiita.imako): like if i make nitroglycerin and then hit it with a hammer it will for sure go KABOOOOM
[13:29] Particle Physicist Bejiita (bejiita.imako): those kind of things
[13:29] Particle Physicist Bejiita (bejiita.imako): certain effects
[13:29] Particle Physicist Bejiita (bejiita.imako): based on certain actions
[13:29] Neuro Wonder: right, it might not know how hard you hit it, but it can ascertain based on what the average hammer hitting force is for the average human (it can look at you and better decide), then, yes, kablooey
[13:29] herman Bergson: Well...enough to think about... :-)
[13:30] Max Chatnoir: Like sometimes female snakes will have babies after years of living in isolation with no males.
[13:30] Max Chatnoir: doesn
[13:31] Max Chatnoir: t mean that fertilization isn't usually involved.
[13:31] Particle Physicist Bejiita (bejiita.imako): hmm
[13:31] Particle Physicist Bejiita (bejiita.imako): virgin snakes?
[13:31] Particle Physicist Bejiita (bejiita.imako): like Maria and Jesus?
[13:31] herman Bergson: I have th think about what it means when AI suggests better procedures....epistemologically
[13:31] bergfrau Apfelbaum: grins Bejiita
[13:31] herman Bergson: What does that imply?
[13:31] Max Chatnoir: yes. All of the babies are male, because of snake sex chromosomes.
[13:32] Particle Physicist Bejiita (bejiita.imako): well snakes work different then humans
[13:32] Neuro Wonder: not quite.....in humans, for example, the sperm is technically not needed to create new human babies
[13:32] Neuro Wonder: and some other creatures can become hermaphrodites when needed or certain stresses prompt it
[13:33] Max Chatnoir: But jesus had to get a Y chromosome somewhere....
[13:33] herman Bergson: Yesterday I sw a documetary about trout....
[13:33] Neuro Wonder: Jesus is written in a book written by humans
[13:33] Particle Physicist Bejiita (bejiita.imako): well unless you have sex with a woman or introduce it by some other menas she does not become pregnant, what i know its the only way for humans
[13:33] herman Bergson: all trout in production are tripoid...infertile
[13:34] herman Bergson: not male nor female
[13:34] Particle Physicist Bejiita (bejiita.imako): ok
[13:34] herman Bergson: males are too agggressive so they want only female trout....
[13:34] Max Chatnoir: Interesting, Herman!
[13:34] Neuro Wonder: the female egg has the genetic material to make a new human with another female egg....it doesn't happen through copulation
[13:34] Max Chatnoir: But not a male.
[13:34] herman Bergson: and they achieve that by mainpulating the eggs...they then all become female eggs
[13:35] Neuro Wonder: unless you have a female with XXY chromosomes
[13:35] Particle Physicist Bejiita (bejiita.imako): hmm ok, never heard of this happening ever though, other then in the story of Jesus and Maria wich is pire fiction
[13:35] herman Bergson: so male eggs are converted into female eggs
[13:35] Max Chatnoir: XXy humans are male
[13:35] Particle Physicist Bejiita (bejiita.imako): pure
[13:36] Neuro Wonder: it's using science and human intervention, nothing that occurs on its own
[13:36] herman Bergson: right..
[13:36] herman Bergson: cause and effect..:-)
[13:36] Particle Physicist Bejiita (bejiita.imako): ah
[13:36] Proxy Badger: What would be the operational definition for intervention?
[13:37] Neuro Wonder: there isn't one
[13:37] Neuro Wonder: using it colloquially because that's how we are speaking
[13:37] Neuro Wonder: basically petri dishes
[13:37] herman Bergson: operational definition?
[13:38] herman Bergson: Actively intervening with the course of events
[13:38] Neuro Wonder: inserting a penis into a vagina is actively intervening
[13:38] Neuro Wonder: but i was referring to labwork
[13:39] Particle Physicist Bejiita (bejiita.imako): yes, and i think this is why it works that way, if a woman could become pregnant by herself it might occur randomly without control but here u need an active action to take place between 2 people or animals who have decided somehow they want to have a kid
[13:40] Particle Physicist Bejiita (bejiita.imako): i think thats the reason behind sexual reproduction, to give control
[13:40] Particle Physicist Bejiita (bejiita.imako): or i dont know
[13:40] herman Bergson smiles
[13:41] Neuro Wonder: if we wanted to, we could get rid of males entirely and the human species would survive. that's why we males better be nice and have a sense of humor and wit
[13:41] Particle Physicist Bejiita (bejiita.imako): it works same even in the plant world with pollination
[13:41] herman Bergson: We now only need to discuss abortion to make full circle...
[13:41] Particle Physicist Bejiita (bejiita.imako): although some plants can be self pollinating
[13:41] herman Bergson: But let's keep that for another day
[13:41] Neuro Wonder: we have to cover death before abortion....
[13:41] herman Bergson: I guess we are off topic enought to end our deliberations :-)
[13:41] Max Chatnoir: Because the flowers have boy bits and girl bits. :-)
[13:41] Particle Physicist Bejiita (bejiita.imako): hmm the not so nice parts
[13:42] herman Bergson: So thank you all again...
[13:42] Particle Physicist Bejiita (bejiita.imako): but a good one again
[13:42] herman Bergson: Class dismissed.....
[13:42] Particle Physicist Bejiita (bejiita.imako): cause and effect indeed!
[13:42] Particle Physicist Bejiita (bejiita.imako): ㋡
[13:42] Max Chatnoir: Sorry I brought up the snakes. :-)
[13:42] .: Beertje :. (beertje.beaumont): ツ
[13:42] Particle Physicist Bejiita (bejiita.imako): lol
[13:42] Neuro Wonder: he's got stamen, she's got style....for the plant enthusiasts
[13:42] herman Bergson: they sneaked in Max :-)
[13:43] Particle Physicist Bejiita (bejiita.imako): they are valid in this case of cause and effects
[13:43] Particle Physicist Bejiita (bejiita.imako): lol
[13:43] Max Chatnoir: Hi, Nicky!
[13:43] Particle Physicist Bejiita (bejiita.imako): maybbee ill program some Python later, i need to get more fluent in the Godot game engine also (uses Python syntax)
[13:43] Nicky (nickyparis.bellecoeur): hiya
[13:43] Particle Physicist Bejiita (bejiita.imako): hi Nicky
[13:43] Neuro Wonder: hi nicky
[13:43] herman Bergson: Hi NIcky :-)
[13:45] bergfrau Apfelbaum: thank you Herman and class! it was interesting again
[13:45] .: Beertje :. (beertje.beaumont): yes very intresting
[13:45] Particle Physicist Bejiita (bejiita.imako): yes indeed
No comments:
Post a Comment