Deepfakes, Chatbots, and Power

For those of you who have been living in a cave without computers or smartphones for the past year, “deepfakes” are videos that have been altered through software to show something that isn’t real or doesn’t really exist. And “chatbots” are a new technology that is taking the internet by storm, that allows software to write convincing text that seems like it is authored by a real person but isn’t. Deepfakes have been around for a while, and have been used politically to, for example, show Nancy Pelosi seeming to be drunkenly slurring her words. Chatbots are new—at least the ones good enough to fool people—and have already struck the fear of God into writers that make their living writing for the internet, or college professors who fear that they will no longer be able to spot plagiarism in their students.
I am old enough to remember a world without computers, and I also remember a time when the earliest forms of social media existed as text posted in DOS on a black screen. Back then no-one could imagine (because the technology hadn’t been invented) that what you were reading was not the work of the purported poster. Then came the age of Facebook and Twitter style social media, where it became possible to “flame” or defame people by posting insults or untruths about a person. Still, there remained a sense that the words posted were the work of a human writer.
Now we have chatbots and deepfakes that remove the certainty of authorship, perhaps forever. Undoubtedly there will be counteractive software that will purport to detect when something is real or fake, but the fakery of chatbots will also continue to improve. I had a long career as a software developer, so I understand how chatbots work. They seem to be talking like a real person, but actually they are consulting billions of pages of preloaded web content and are actually only programmed to pick the next word, whatever that is. They work a lot like computer chess programs, consulting billions of chess games to compute the next move. These chess playing computers are now good enough to beat world champion human players.
To me the real issue is not technology, or the continuing weirdness of the post-truth world we now live in. To me the real issue is the age-old exercise of power. These technologies, like drones, like swords, like semi-automatic rifles, are or can be instruments for the wielding of power, and power is perennial. To understand power in the human realm, I think we need to remember that human beings are social animals, and in particular primate animals. Tribes of human beings behave much like tribes of chimpanzees or gorillas, with whom we share 98% or so of our genes. Among social animals, males are constantly vying for supremacy, and there is nearly always an “alpha” male, who has power over all the other males, as well as mating rights with the females. This makes some kind of evolutionary sense, if we think that the best way to transmit the strongest genes to the next generation is to have the strongest male impregnate all the other females.
What I see coming is that these new internet fakery tools will become just another way for people (probably preponderantly male people) to wreak their will on everyone else. I’m sure that even now adversarial nation states are working on ways to further undermine our social and political fabric by flooding the internet with deepfake and chatbot content so good as to convince all but the most wary. I say “male people” because of the chimp factor. Why is it that we so rarely read of a female mass murderers? The data shows that female shooters are exceedingly rare. I’m no social biologist, but it seems to me that this is related to our primate ancestry. It is the males who seek power, and will use any tool available, whether a weapon you hold in your hand, or a weaponized technology such as deepfakes or chatbots.
Here’s a living example. One of the ways deepfakes are causing great suffering is a particular kind of revenge porn, perpetrated by males (often ex-boyfriends) who use deepfake software to modify an existing porn video to show the face of an innocent woman. Many women’s lives have been ruined by this weaponization of video technology; once the video has been posted there is really no remedy.
I wonder if before too long there will come a time when face-to-face human contact—reduced so much during Covid—will have a renaissance as the only way to insure that the person communicating with you is a real person and are who they say they are. I believe that the origin of the handshake as a greeting custom came from this kind of thinking. By extending an empty right hand you are showing that you do not hold a weapon, and can be trusted not to attack or kill. I can even see sci-fi novels being written about such a world, in which computers are so mistrusted and so weaponized that only real human contact and real human touch is trustworthy. The old becomes new again.
Final question: how do you know that I have in fact written this blog post, and that it has not been written by a chatbot? A chatbot can be loaded with all the books, blog posts and internet content I have ever written, and programmed to imitate my style of writing. At the moment you probably trust that the writer here is me, but watch out. Chatbots are AI software, which means they are programmed to continuously learn. Every day they are getting better. One day they might be good enough to imitate YOU.