Deepfakes are a thing now. They are simply really good fakes. Nowadays many of the videos of politicians are deepfakes. Did you see the videos of Tom Cruise earlier this year on TikTok? They were deepfakes. Those who watched the documentary recently about the traveling chef Anthony Bourdain might know that the editors inserted a deepfake sound bite in Bourdain’s signature voice, which got them into hot water, deservedly so. Pretty soon we won’t be able hear any difference.
Deepfakes are now focused-on video and audio. It’s getting to the point where pretty soon you won’t be able to detect any difference from the original. What about fake news then?
The current doyen of video is clearly TikTok. It’s not clear to what extent deepfakes have penetrated it also but it would be foolish to discount the possibility. But there’s an intriguing angle here. That’s the recent approach by TikTok to encourage its users to do TikTok video resumes. That’s a new one.
Actually, it does make sense, you can see a lot more in a video resume than you see on the traditional text-based form. But now you have got to contend with the possibility that the all-singing all-dancing engaging resume that just entertained you no end was in fact a deepfake. Bummer, she would have been a great hire.
There’s probably a limit to what you can do with a deepfake video resume based on common sense and normal skepticism. An introvert will strain to be a credible extrovert and vice versa although it can certainly be done. But what if we breach this barrier?
The heart and soul of TikTok is its recommendation engine comprising algorithms that categorize its users by the sort of videos they want to watch. This obviously is highly proprietary and was one of the reasons the Chinese company dug its heels in when the last President attempted to have them sell themselves to an American company.
But what we do know is that TikTok observes your reactions minutely when you view different types of videos. It uses what the economists call revealed preference to find what you really like and don’t by looking at how long you linger on certain types of video content. It’s all about eyeball persistence.
What TikTok is actually doing is a sophisticated test of personality on the user, way more advanced than the traditional tests like the Myers-Briggs. We don’t know what’s inside the engine, but it probably can break down a user’s interests into millions of categories and improve the accuracy of the personality test a million-fold. When it’s done it knows the user better than the user herself. In other words, you can really see TikTok as being a personality engine as much as a video engine, and its real claim to fame is its determination of your micro-personality.
So, where’s all this going? TikTok has the goods on you. It knows what you like and don’t to a million levels of detail. It has the psychometric content on you to be able to mess with how you present your personality. It now has the wherewithal to change the way you express your personality – with or without your approval.
In other words, TikTok has the means to create a synthetic personality. Maybe it will allow you to do this explicitly and formally and if so, well and good, maybe. If not, who knows? That engaging video resume can still now be amped up to Michael Jackson and Barbara Streisand heights. How would you know that if you’re just seeing the person for the first time? And if you already know them, are they just having a very good day, or are high on something or the other?
Now I’m just surmising about TikTok but if they don’t do it, others will, and TikTok don’t appear to be sluggards to me. Chances are the horse is already out of the stable.
Of course, video resumes are just one possible use of deepfaking and synthetic personality. Video games are another. How cool would it be if you understood the personality of your videogame opponents and so you know what they like and don’t like even before they do? If you’re playing a simulation game, great to be able to guess what he will aim for before he actually does. War games. Dating apps. You get the idea.
We’ve been cocking our collective snoots at the wrong food bin. The literati have been worried about deepfake videos and audios. But that’s just the kiddy stuff. The real target is being able to build synthetic personalities for app users on the fly and then apply it to situations where the user, and more likely the app vendor, can use it for their own purposes, altruistic, venal or otherwise.
The real aim of deepfakes is not a better video or voice, it’s a better understanding of your personality, your vulnerabilities and weaknesses so that they can be exploited for whatever style of online manipulation comes into vogue next.
We have to realize that this hasn’t just appeared all of a sudden. I last wrote about this evolution in 2014
(“Can Siri Become a Great Leader? – The Coming Out of Synthetic Psychology, 11 January 2014”). My point then was to highlight what I saw as the possible evolution of digital assistants like Siri (remember her?). There wasn’t any TikTok then, but the game has now been upended by the growth of video, Zoom, Alexa and other video apps.
Synthetic personality is just about here now. Pretty soon you’ll be able to select a personality for yourself when you Zoom, video resume, Twitter, Instagram, in fact whichever way you choose to communicate, even if it is not by video. You will be able to make a choice or choices; a bit like Charlie Sheen, a lot like Ann Nicole Smith, more than a bit like Tom Cruise. You’ll be able to choose the degree of charisma, gravitas, visionary-ness, inspirationalityy and so on for whatever you want to portray.
The Holy Grail is the selectable personality. There will be millions of templates for you to choose from (“I want a mixture of Michael Jackson and Steve Jobs, 60:40 thanks, just hold the dancing bits”). For dating apps, hold the creepy bits. For resumes, amp up the gravitas, equal to an age of 52. Etc.
The 64-dollar question is, will you be able to patent something here? The personality categorization process perhaps? Or the personality templates that result? Can you patent the behavior of the models and users from whom the template derives? How about the facial recognition process that teaches itself how to categorize the faces of different personalities?
Personality is identity. Can someone therefore patent identity, or some cognate (like legally responsive behavior pattern?) Is TikTok actually appropriating the identity of its users? What then?
Will we be able to buy a new personality, one that is more to our rapidly-changing tastes?
I’ll have a dozen like that, please.
When you subscribe to the blog, we will send you an e-mail when there are new updates on the site so you wouldn't miss them.