Julia Galef - 2016-03-02
(ETA: Changed the title, since it was misleading!) How is rationality like artificial intelligence? One connection is that both fields are interested in how to handle interdependent beliefs. In this video I explain why your brain is *not* like an AI, and why that means you end up believing contradictory things.
I like your use of the Penrose triangle
+Brad Younger The example was so good. Made me smile.
I think some people do look for orphan beliefs. One of my friends is said to be "slow". But he graduated in math with high scores, so that seems empirically false, specially since undergrads in his faculty were given very little time to do each exam. The reason they call him that, I believe, is that whenever he is given any new piece of information, he tries to first figure out exactly what the information means and then fit it into his other ideas. If this is so, some people will look dumb while actually being more rational than others.
I'm suddenly seeing a relationship between the Penrose Triangle and Cognitive Dissonance.
I like the Penrose triangle analogy. I have always thought that a person should continually broaden their horizons so as to avoid getting stuck in one parochial corner of the triangle.
Absolutely! That should be one's goal always, at least in whatever area of expertise you choose. Of course, applying that to a broader reality such as life itself makes it more interesting. Just last night I was watching a film related to AI, Ghost in a Shell (anime) the newer Hollywood movie misses a few concepts or at least the way they cover them doesn't feel the same.
1:02 "Normative" and "Ideal" are not words I ever expected to hear in the same definition, outside of a lecture on neoclassical economics, that is. 😂
i was really missing your videos.
This explains why I still avoid eating pork and shellfish, even though I am no longer a member of a church which observes the Old Testament dietary laws.
You get some harsh criticism :) This is a great insight expressed very clearly. Thank you!
Thanks :)
Amazing - thank you for your efforts is educating and sharing Julia !
That backdrop is really confusing the color temperature.
looks normal to me
she must have selected the auto color correction in her camera settings, hence the confusion.
set the auto color correction to manual and know what color your light is producing and set it to that level.
Do not attempt to adjust the picture. Julia Galef is controlling the transmission.
LOL
love your work, would love to collaborate on something sometime. currently researching a high level series on AI/Cognitive Science
3:11 In other words they are dependant variables.
Thank you so much! Great ideas explained in a very clear and followable way
Hi! Thanks for your videos. Do you have a good tutorial / reference to understand and use bayes nets? I would really appreciate it (Specially with python)
Thanks, nice lucid explanation of a Bayes net. This is the first time I've heard of "orphan beliefs," too.
There is also a flip side to this, if I understand it correctly, of being careful not to use this line of thinking to commit the genetic fallacy http://lesswrong.com/lw/s3/the_genetic_fallacy/.
Put more concretely, when you discover that a belief is "orphaned" because you no longer hold its parent belief, this is just evidence (however strong) that the orphaned belief is not true, not an excuse to dismiss it outright.
Indeed. Good point.
Interesting. Yeah, these emotional responses tend to suggest reactions in the future even if your logic remembers that this was not actually the case. Referring to the hygienic case.
Love your videos! This one is particularly relevant to my realization that free will is a joke. I've spent the several decades since that time looking for and altering "orphan beliefs"—great term—which has made a huge difference in my happiness.
The blue hurts my eyes. It seems to fluctuate a bit, and its just too bright. Maybe tone it down somehow. Darker color, less light on the background.
i dont, im perfect.
wait, there was a backdrop? I was so memorized by what she was saying I didn't even realize there was a background.
According to that example, I am defenetely an AI implanted into a bioloical container X)
This is fantastic stuff! Liked and subscribed.
I’m new to your channel. Just saw your Ted talk while looking for new speakers on neuroscience of cognitive biases.
Are you collaborating with anyone in neuroscience to identify where exactly in the brain the orphaned beliefs take hold and why they stay ingrained for some, while others can acknowledge and remove them?
I look forward to learning more about your work, especially the application of Bayes for the public.
Very inspiring!
Could you also name your favorite books on statistics and others that has helped you become such a rational geek? :)
I forgot how much I needed this. These videos are very intellectually stimulating.
I absolutely love your videos! Keep them coming!
My area of research is misinformation, and it resonates a lot with what I'm doing. You might be interested in looking up about the continued influence effect. When you talk about the hand sanitizer, that seems to be what happened. Even if the misinformation was corrected and acknowledged, you continued to act as if it was still valid. This has been abundantly studied in psychology :)
Johnson, H. M., & Seifert, C. M. (1994). Sources of the continued influence effect: When misinformation in memory affects later inferences. Journal Of Experimental Psychology: Learning, Memory, And Cognition, 20(6), 1420-1436.
Thanks for sharing this! I love how simply knowing the right term makes all the difference between google not giving any results and google full of results on the subject that you want.
Thanks your idea is really helpful for a new model I am building :)
Bayes nets are just one of the minions of AI. But still, I love your insights. I like you too! I can't help it am sapiosexual and you're cute
Thank you so much!
Hmm. Maybe, I'm an AI. I do tend to update all of my related beliefs / assumptions when new data comes in.
Sure, I might miss a connection here and there, but my world view adapts to evidence.
just like when you google: why my internet not connecting.
Who is this Genius Wise Beautiful Mature Attractive Aware & Lovely Women!?
Thank you for this video. I subscribed because of Science4All
That last bit with the Penrose triangle gave me a much more conscious appreciation of the work of M. C. Escher. Thanks.
Nice video, thanks. I'm curious as how this fits with Stanislas Dehaene's theories: http://www.college-de-france.fr/site/en-stanislas-dehaene/course-2013-01-22-09h30.htm
Brilliant video! Very well explained, and you chose a very relatable experience to share.
你真好看.
Coucou Julia..Ça va ? Your vidéo is interesting. Perhaps I could use you logic to develop composite Key performance indicators if you could help me :)
This is really good. I feel like you skipped evolutionary psychology in your explanation and could have explained how the brain works and how we solve problems automatically.
PLEASE UPLOAD MORE
i'm so confused, are you aubrey plaza or not?
Just found your channel, first video I have watched, and will subscribe, so forgive this if you have dealt with this issue in previous videos, I found Bayes in graduate school, and have have generally found it a good tool in rational thinking. But, there is a seductive weakness: when an individual assigns a probability to new, incoming evidence (evidence whose probability is not objectively quantifiable or unknown) then that subjectively assigned probability is subject to confirmation bias. The individual, familiar with the theorem, knows that even minor changes in the probability of subsequent evidence can greatly affect the ultimate probability, may subconsciously skew their evaluation of the probability of new event, even slightly, in favor of the outcome that they believe in, even if that belief is totally sub-conscious, thus crating an incorrect conclusion, but one that will fit their preconceived conclusions. How so we deal with this?
neat stuff. :)
I look forward to your very well made videos.
Ok, first of all, very glad to know you are back. I was missing these videos. They are really great and I enjoy them a lot.
The notion that human belief inconsistencies are due to antagonistic orphan beliefs is very interesting. But I was thinking that maybe propagation does occur, and the cause for inconsistency perpetuation is another one:
What about the notion that our consciousness may not be based in a single "network", but on a set of networks (maybe similar, but independent), and at any given time, given the current environment, current emotional state, current awarness, etc.. a particular network is activated; So, for instance, even if I had updated a belief, it may be updated in a particular network, that in a particular state may not be activated, and the one that is active is not updated. Then, if I realize this, I can update this network consciously and carry on with a more consistent set (that may need furhter updating on other elementary networks).
I guess what I'm trying to say is that propagation does occur very efficiently on a given network, even for humans. In the religion/monogamy example the key assumption is that monogamy was a child node of the religion-node and had no other parents. I think a belief in monogamy can be preserved easily if an empathy-node (independent from religion) also parents the monogamy node. Also, the networks in human minds (I figure) are pretty dynamic, so instead of deleting the religion-node and leaving orphan nodes, I woul think that what actually happens is that reason-based-morality/integrity node(s) replace the religion-node...
Maybe my brain is a set of dynamic bayes nets, and if something like the could be implemented in A.I. then maybe there would be no difference at all.
too good content and brilliant thinker!
It may be the case, that misinterpretation of the article was influenced by your tension around the way your team were promoting sanitizers (and unrecognized by that moment). So, obtaining of the correct interpretation left that tension in more verbalized form.
hasen195 - 2016-03-02
With orphaned beliefs they don't really get orphaned. Once an idea is planted, it will find other supporting nodes, so that even when its' disconnected from its original node, it still has firm roots in the mind, specially if it's been there for a long time.
Isaac Abodunrin - 2016-03-02
+hasen195 I totally agree with you; I notice I tend to rationalize/justify my orphan beliefs subconsciously.
hunglikehuang - 2018-04-07
Bingo, especially given our tendency for confirmation bias.
Rogerio Santos - 2019-11-24
Also, as one conforms to the orphan believe before it being orphaned, one needs to justify this behavior, like a chastity life, to use her exemple and not seem stupid, thus keeping the "I'm not stupid" believe and avoiding cognitive dissonance. The problem was to think about the believes with only one tie to another believe, when it grows a net the more one thinks about it. In this exemple, there could be believes about STDs risk, shallowness of relationships, lost of social value (actually very true to women), risk of violence etc. Those beliefs don't go away with the religious one.