The Ethics of Self-Generated Profile Pictures
And a few more tools for the Synthetic You
Take a look at this picture of me. . .What do you think?
Am I beautiful? 😄
If the title of this post didn’t already give it away, you should know that this is an AI-generated image.
What if I made it my new profile picture? Would that be weird? Dishonest, maybe?
Would you ever use a generated picture of yourself as if it were the “real” you?
I think at the core of this dilemma is the question of whether an AI generated profile picture is a fair representation of the person it portrays.
Well, what is a “fair representation” of you, anyway?
Who you are fluctuates across any period of time—or at least, how you express yourself changes moment by moment.
You are at times happy, sad, mad, disgusted, surprised, and scared.
But I imagine your professional picture shows contentment and confidence.
Are you always perfectly groomed and dressed to impress? Is the emotion that your profile picture portrays the emotion that others will perceive 95% of the time while being in your presence? Is that expression what the people closest to you—who “know” you the best—experience in their typical interaction with you?
Perhaps you could say that there is a context in play where everyone understands and expects that your picture will be one of the better "slices" of you and that no one understands it to be holistically representative.
Fair enough. . .But then couldn’t you eventually make this same cultural argument for AI?
Provided the new guard is able to subvert the old guard, we would inevitably arrive at a new cultural equilibrium, where the same principle of approximate representation is in effect, albeit to a novel degree.
Once a critical mass of people employs self-generated profile pictures, the shock factor will wear off, and expectations will be recalibrated accordingly.
I’m not saying that this would be a good outcome. Rather, I’m suggesting that the base element of this dynamic already exists and is accepted as the default norm.
In essence, AI generated pictures of yourself are merely an adjacent step along that same spectrum of approximate representation. Of course, it's worth thinking about whether there’s a point that is objectively “too far” along that spectrum, as well as identifying on what basis (independent of cultural inertia) we could even make such a claim of ‘too far’.
Maybe you could claim that with AI generated photos there is some residual of inauthenticity compared to a photo taken with a camera. You might even say that though your current profile picture came from a point-in-time snapshot, at least it was a live capture of a real moment that actually happened within the timeline and physical space of human life.
Yet even with a camera, there is still some “generative process” at work, as the semiconductors digitally encode and arrange the pixels along an image file, such that we accept that the image produced by the camera “fairly represents” reality.
And beyond that, most people are comfortable with the subsequent layer of after effects: balancing out lighting, improving skin, and other aesthetic-enhancing adjustments.
So at the highest level, this is all part of the same issue. Do you think that AI-generated profile pictures are a step beyond the ethical line of what should be acceptable?
—Drago
P.S. I was recently granted access to another cool AI tool that can detect emotion within facial expressions, tone of voice, and choice of wording. It’s pretty amazing.
Check out how it decomposes my expression into different factors:
P.S.S. Any technological tool takes in input and spits out output. GPT has nearly perfected output. So how to innovate going forward? To develop a truly innovative product with GPT, the creative challenge is to identify ways to optimize the input—the data collection process and the manner of organizing that data (converting it to a nice prompt that is digestible for the language learning model).
I.e. right now the steps of interacting with GPT are:
Have some idea of what you want to do.
Have an idea of the kind of question you need to ask in order to get the “advice” from GPT.
Structure the question/prompt such that the answer GPT provides will be fully optimized.
Physically type the prompt into the chatbox and then click “Enter”.
The way you can create value in the GPT Economy is to innovate within these four steps.
This won’t be an easy task, however, as you may have seen the announcement about GPT Plugins . These plugins, despite the immense benefits they provide to you as a consumer, will make it even more difficult as an entrepreneur to independently penetrate this space.