What About Deepfakes?

As technology advances, it is only becoming easier to lie. That statement might be surprising at first glance. With the advent of photography, audio, and video recording, we have gained access to more truth than ever before. We are now able to hear and see events that we were never present at—events that happened decades ago and thousands of miles away. Surely this should only make it more difficult to lie!
AI Advances
Photos and videos, however, are the perfect media for communicating half-truths. First, they can be altered. And even if they haven’t been altered, they are still only a narrow perspective, a sliver of “what really happened.” They’re an outline from which our mind colors in further details. We are more easily swayed by visuals than by text, because we are used to having other people lie to us, but we are not used to having our own senses lie to us. That is the power of manipulated media.
With better technology, our lies have become more convincing. For almost as long as people have been taking photographs, people have been falsifying photos. The same goes for audio and video. With Photoshop, anyone can paste one person’s head onto the body of another. With video editing software, anyone can slap a different audio track on top of a blurry video. Or, with a little more effort, splice together a parody of President Obama’s State of the Union address with sound (and video) bites taken out of context.
Artificial intelligence (AI), and in particular “deepfake” technology, accomplish all this with far more sophistication. A deepfake, which is a portmanteau of “deep learning” and, well, “fake,” is an AI-generated image or video of anything you want. There was an amusing deepfake of Pope Francis wearing sunglasses and a puffer jacket (the “Balenciaga Pope”) which circulated the internet in early 2023. Even this seemingly innocent example deceived many people, in part because it was unprecedented.
The uses of deepfake technology only get darker from there: the aim of the earliest software was to generate falsified pornography “starring” various public figures (and whomever you liked). And now, generating a convincing deepfake of a State of the Union address is easier than learning how to use video editing software: just tell the AI what you want the president to say and twiddle your thumbs while it mints an original video of the “event.” In fact, don’t write the transcript yourself. ChatGPT can do a better job. We no longer need half-truths—or human imagination—to tell a convincing lie.
Read More»

Jack VanDrunen | “A Christian’s Take on Deepfakes” | January 2024 Issue of New Horizons


RESOURCES

Heidelberg Reformation Association
1637 E. Valley Parkway #391
Escondido CA 92027
USA
The HRA is a 501(c)(3) non-profit organization.


Subscribe to the Heidelblog today!


Comment

Your email address will not be published. Required fields are marked *

Comments are welcome but must observe the moral law. Comments that are profane, deny the gospel, advance positions contrary to the Reformed confession, or irritate the management are subject to deletion. Anonymous comments, posted without permission, are forbidden.

This site uses Akismet to reduce spam. Learn how your comment data is processed.