Wonderful article in the November 20 New Yorker about deepfake technology. In it, Daniel Immerwahr notes that although the technology exists to create convicing videos of people doing things they never did, or saying things they never said, this has yet to have any impact on our politics.
First of all, almost all the high-end deepfake videos made so far are porn: "Able to depict anything imaginable, people just want to see famous women having sex." (Asked if she were going to pursue legal action to get such vidoes of herself off the internet, Scarlett Johansson said she wouldn't bother, because "the Internet is a vast wormhole of darkness that eats itself.")
Immerwahr goes on to question whether fake political videos would have any impact even if they were made. He notes that you don't need video to convince people of false things. There was never any evidence that Catherine the Great had sex with a horse, or that Marie Antoinette said "let them eat cake," or that Barack Obama wasn't born in America, or that Covid-19 was fake, or that Trump really won the 2020 election. We repeat such things "because they affirm our fundamental beliefs, not because we've seen convincing videos."
Mind you, it isn't that fake political videos are not being made. They are, all the time, but they are not made to look convincing: "if by 'deepfakes' we mean videos produced using artificial intelligence that actually deceive people, then they barely exist. The fakes aren't deep, and the deeps aren't fake." Instead of realistic videos designed to fool us, we get Dark Brandon shooting lasers out of his eyes. Immerwahr:
There is something gratifying about having your world view reflected visually; it's why we make art. But there is no need for this art to be realistic. In fact, cartoonish memes, symbolically rich and vividly expressive, seem better suited for the task than reality-conforming deepfakes.
Fretting over new kinds of evidence gets us once again back to the fundamental problem that most of our beliefs seem radically undermotivated. Why we believe things is a very hard problem, and sometimes evidence of any kind seems to have little part in the process. That, not deepfake videos, is our real problem.
11 comments:
I wonder if "beliefs" are really much about evidence at all. One holds *and* voices beliefs, it seems to me, for social, psychological, endocrinological, and host of similar/related reasons.
Meanwhile, the informal use of evidence in everyday contexts (hot stove=danger), on the one hand, and the rigorous use of evidence to draw articulated conclusions, on the other, may be effectively distinct cognitive functions. The latter usually/often reflects professional training and work activity, and isn't a source of relaxation or sociality for many people. Many don't bring that kind of rigor to say, discussing things with friends, posting things online, voting, charitable giving, the things that make them worry or avoidant, etc. (Plenty of exceptions here; see just below, but also nerdy fandom discussions, recreational machine work or crafts, chess, etc., etc., etc.)
It's true there's a community of people like Scott Siskind who come out of Less Wrong and have made a study of using rational, evidence-based rigor in their regular lives. They've read Yudkowski's _Sequences_, and so on. And there have probably been similar movements in the past. And there are individuals (like, dare I say, yourself) who aspire to the same, without cleaving to a movement.
This sort of practice in non-professional contexts is probably alien to most people, and alien to what they want out of beliefs and their expression; hence the hostility evoked by EA, for example, which Siskind writes about a lot. And yes, some of this discomfort on the part of the mass (in which on this point I would include myself, although I don't have strong feelings--note the word--on EA) is about feelings of inferiority and discomfort with difference and is basically discreditable.
But maybe I've strayed from the real point. People have beliefs for the range of non-evidence-based reasons I've given. You could have all the evidence in the world--ALL the evidence. Sometimes it will make a difference: presented with evidence in the right socio-psychological way about something not deeply important to them, they may say, "hey, you're right." Sometimes, if it's important to a person, the evidence won't matter. "You know, X politician is just my kind of leader, and that's all there is to it." "You know, I don't want to live in a world of stoic reticence, and that's all there is to it." "I believe God is with me, and nothing you say can shake that, because I feel it deep down" (thinking of Wm. James' examples). Etc., etc.
Perhaps there is room here also for the suggestion, which got some press a few years ago, that discursive rationality, if that is the right phrase, evolved as a sort of force multiplier in social relations. That is, it evolved so its bearers could win arguments and gain social clout. This is, of course, the position of the sophists. Yours is the Socratic position. Alas, I think most of us are sophists.
To be fair to the sophists, if this theory is right, then the whole quest to use reason to find truth is a bit jury-rigged, no? It's like land vertebrates using the throat, which fish used for alimentary functions, for the completely separate and often incompatible function of breathing. (I'm fascinated by this observation, which I owe to Scott Atran's _In Gods We Trust_, which has much to say on cognition, the evolutionary value of holding counterintuitive and counter-evidentiary beliefs, etc., etc.)
Then there's the question (heh, I could go on all day) of how different, really, are even the most powerful users of reason and evidence from the rest of us, at least when it comes to social, psychological, moral, etc., type issues. Marx was unquestionably a powerful reasoner, and he gathered masses of evidence. And yet he always, always, always came up with, well, more Marxism. Did NONE of that suggest something else? Ludwig von Mieses the same, for the opposite camp. Doesn't that pairing suggest that reason and evidence were not the only things at work in these two brilliant men? No, I haven't read even an important fraction of Marx, and I know almost nothing about Mieses. I wait to be corrected by, um, evidence. I still think my point is a valid one, and that's all there is to it.
The question of how much impact evidence has on our big discourses is, I think, one of the most important. I am flattered to be told I am on the side of Socrates, because I do try to let reality change my mind.
Sometimes it seems to me that facts don't matter at all, as with young communists or rigorous libertarians. But can that be so? How could we have built this astonishing civilization if we are not engaging with reality in a serious way?
I suppose we must work in different modes; imagine a far right ideologue who is also a great motorcycle repairman, or a communist veternarian.
Since World War II, the US and western Europe have been able to keep democratic politics and freely-wheeling culture attached strongly enough to reality to keep the wheels of progress turning. But was that a fluke, or something we can expect to continue?
What can we do to help maintain it?
On your last question, I have no answer. Sometimes I think some rough beast is slouching toward us, a new civilization I will barely understand--and sometimes I wonder if that's just what people my age do. Incidentally, I'm not *just* thinking of Trumpism here.
Letting reality change one's mind is good. But perceiving reality is very, very hard, largely because of confirmation bias, which I realized afterward is what I was trying allude to with my Marx-Mieses just-so story. We all engage in it. Our biases and leanings are very deep in us. They are our selves. (If the self is an illusion, what is it that is deceived? How would that work?)
For my insistence on the fundamental importance of some kind of internal quasi-reified psychological THING which either shapes or is the same as our deep preferences, I lean on William James, whose _Varieties_ depends heavily on such a concept and the patterned varieties of it across individuals (in fact, those seem to be what he means by The "Varieties" of Religious Experience). If it's good enough for WJ, I'm good enough to use it too.
I would add that I'm deeply skeptical of the value of reason and evidence in the kinds of conflicts we have now, which are about fundamental values (i.e., biases=psychology=selves). With some of our conflicts, what would that even look like? "A methodology-heavy survey of the evidence shows that countries without an established religion have, on average, a GDP 18.5% higher, and to score 10.2% higher on Nimbu's Weighted Happiness Scale, than countries with an established religion, when one corrects for differential development parameters." Is that going to get Mike Johnson to suddenly exclaim, "Whoa! I've been wrong all this time!"--with subsequent euphoric hopping about, like Scrooge on Christmas Day? No--it'll just sound like confirmation bias to him. And will he be wrong?
Of course, a survey like that is really Reason n' Evidence Lite (tm), isn't it? It's humans performing reason n' evidence. Maybe Real Reason And Evidence are totally different. Bring on the AI!
During last few months (mind you, first before the election and then directly afterwards) two deepfakes were created, were an AI-created and synced fake audio for real politician videos. Those two wre created by two opposite parties. Both were marked as fake - first, with a footnote-like remark "audio created by AI lector" and IIRC was based on most likely real transcripts and second was totally fake, but clearly marked as such. Despite that I saw numerous people commenting right under those fakes seemingly thinking they were real.
BUT
I saw numerous people thinking other fakes are real and nothing, nothing can convince them otherwise. So yeah, I think this is not something new.
@David- I am not sure that our discourse points in the direction of a unified "self." One might equally say that in points in the direction of muliple selves, some of which are rational and respond to evidence, and some of which do not.
@John
Sure--I think multiple selves will do just as well.
It occurs to me that reason and evidence aren't unified either. Bad reason and bad evidence are still reason and evidence. Conspiracy theories often quite consciously perform reason and evidence--like starting from the chin-stroking question "cui bono?" Figure out who stood to benefit from killing JFK and you'll know who did it, and the evidence will start to fall into place. Well, no. But it's not just because emotion, or the pressure of "the irrational." The Laffer Curve, the Domino Theory, the Oedipus Complex--were these not "rational"?
"Since World War II, the US and western Europe have been able to keep democratic politics and freely-wheeling culture attached strongly enough to reality to keep the wheels of progress turning. But was that a fluke, or something we can expect to continue?"
I think I've been trying to say that the idea that that unified culture was based in reason and evidence, or that its values were in some reason-sound sense attached to reality, is an exaggeration, possibly a myth. I like much in the way of its values, and I'm sad that those values are dying--so dislike of that era and its culture is not my motive.
Humans have indeed gotten very good at absolutely spectacular elaborations of instrumental reason (rooted in hot stove=danger). We can build and repair motorcycles, concoct, practice, and improve veterinary medicine reasonably well, build CERNE and the James Webb telescope, and so on. And we're pretty good at developing professional-context understandings of things like biology, realizing in professional contexts that new knowledge necessitates rethinking, etc.
But, as I keep saying, *values* are not based in reason and evidence. A performative version of reason and evidence may be used to buttress them, but its status as, let's say, professional-quality reason and evidence is pretty thin. This includes democracy, human rights, etc., which, as Yuval Harari points out, are just as much fictions as Babylon's monarchy, libertarianism, communism, or the American consensus c. 1960.
That consensus did (somewhat?) exist, as much as consensus ever exists in complex societies. Rick Perlstein's Nixonland is a brilliant chronicle of how that consensus broke up. And our lines of division today are descended from those he paints, although the contours of those divisions have (again, somewhat?) changed in the process.
That breakup was not (or maybe not just) the result of a failure to use reason and evidence. It's about the still-mysterious processes by which collective human values change on a grand scale.
Post a Comment