Thank-you so much for this piece. I've become the AI-police among (mostly) older friends spreading deepfake video recently - this is a very helpful guide to share with those who are still willing to read in depth.
Thank you so much for this piece. This is so important for kids just coming up to using technology, and us elders who, I have to admit, took a minute to see what is actually going on.
It's a learning curve, but one that can be achieved by any age person.
The visualization of the ladder is extremely helpful.
Decades ago I wrote _Digital Compositing in Depth_ about creating "the art of the invisible effect" and one decade ago my doctoral dissertation _Accountability by Camera_ about public video holding police to account. Every point, every piece of advice you give here is invaluable.
Unfortunately, humans believe the first image we see, and no amount of words can completely remove that visual influence. The tyranny of the visual. The first lie, presented visually, is almost impossible to erase, and the more realistic the more indelible.
Normalize not knowing things 👏 I really like how the ladder reframes “I don’t know yet” as a valid stopping point, not a failure or a gotcha moment. Turning verification into a deliberate pause-and-think (rather than a reflex to decide or share) feels like the right mental model for this moment. Thanks for such a clear and humane framework.
Something one of my mentors (in wildlife tracking, but it's relevant anywhere) pointed out once is how much American culture (others too certainly, but I can speak only for my own) prioritizes having the right answer, even though most of the time, it's not a matter where being right is life or death. The rush to be right makes getting the wrong answer feel like failure, and uncertainty is uncomfortable because the question hasn't been resolved.
The discomfort of uncertainty can be uncomfortable, sure, but it won't kill us. That was the beginning of my journey to reframe being right or wrong not in terms of success or failure, but as part of a process that doesn't really have an endpoint.
Incredibly valuable information here. The whole AI situation is a nightmare on every level. The changes humans have been dealing with for the past 200 years or so have been enormous, destructive and difficult, but I believe this one has the potential to destroy our culture.
I no longer work as an instruction librarian but I am loving the application and extension of principles of information literacy here. Tracing something to a source when I'm not sure has been helpful so many times. I study and practice wildlife tracking, so I see a *lot* of animal videos, many of which are fake. Often it's a matter of recognizing that the behavior being shown isn't realistic, but if a person doesn't have that expertise, what are they to do? It's the same process I used to teach undergraduates: if you can't verify something by the content itself (increasingly difficult with AI generated videos, as you observe), there are these other ways to assess it.
Even though I am savvy, I've been fooled recently. A friend showed me a music video on her television set that she liked. I'd never heard of this supposedly lost blues artist named Ruby Mae Lightning but the song caught my attention. Something seemed off and within minutes we were both astonished to learn through the verification ladder it was an AI-generated recording in heavy rotation on a music platform.
Thank-you so much for this piece. I've become the AI-police among (mostly) older friends spreading deepfake video recently - this is a very helpful guide to share with those who are still willing to read in depth.
I'll be sharing this for those who aren't able to discern reality from AI. I appreciate your wealth of knowledge more than you can know.
Thank you so much for this piece. This is so important for kids just coming up to using technology, and us elders who, I have to admit, took a minute to see what is actually going on.
It's a learning curve, but one that can be achieved by any age person.
The visualization of the ladder is extremely helpful.
Brilliant, thorough, and much needed.
Decades ago I wrote _Digital Compositing in Depth_ about creating "the art of the invisible effect" and one decade ago my doctoral dissertation _Accountability by Camera_ about public video holding police to account. Every point, every piece of advice you give here is invaluable.
Unfortunately, humans believe the first image we see, and no amount of words can completely remove that visual influence. The tyranny of the visual. The first lie, presented visually, is almost impossible to erase, and the more realistic the more indelible.
Normalize not knowing things 👏 I really like how the ladder reframes “I don’t know yet” as a valid stopping point, not a failure or a gotcha moment. Turning verification into a deliberate pause-and-think (rather than a reflex to decide or share) feels like the right mental model for this moment. Thanks for such a clear and humane framework.
Something one of my mentors (in wildlife tracking, but it's relevant anywhere) pointed out once is how much American culture (others too certainly, but I can speak only for my own) prioritizes having the right answer, even though most of the time, it's not a matter where being right is life or death. The rush to be right makes getting the wrong answer feel like failure, and uncertainty is uncomfortable because the question hasn't been resolved.
The discomfort of uncertainty can be uncomfortable, sure, but it won't kill us. That was the beginning of my journey to reframe being right or wrong not in terms of success or failure, but as part of a process that doesn't really have an endpoint.
Incredibly valuable information here. The whole AI situation is a nightmare on every level. The changes humans have been dealing with for the past 200 years or so have been enormous, destructive and difficult, but I believe this one has the potential to destroy our culture.
Thank you for your work.
I no longer work as an instruction librarian but I am loving the application and extension of principles of information literacy here. Tracing something to a source when I'm not sure has been helpful so many times. I study and practice wildlife tracking, so I see a *lot* of animal videos, many of which are fake. Often it's a matter of recognizing that the behavior being shown isn't realistic, but if a person doesn't have that expertise, what are they to do? It's the same process I used to teach undergraduates: if you can't verify something by the content itself (increasingly difficult with AI generated videos, as you observe), there are these other ways to assess it.
Great instructions!
Even though I am savvy, I've been fooled recently. A friend showed me a music video on her television set that she liked. I'd never heard of this supposedly lost blues artist named Ruby Mae Lightning but the song caught my attention. Something seemed off and within minutes we were both astonished to learn through the verification ladder it was an AI-generated recording in heavy rotation on a music platform.