I don’t use ChatGpt. Or any AI tools.
It’s a bit of an ethical dilema for psychologists.
Where is the data going?
Who uses that data and for what end?
What happens when it hallucinates and makes things up?
As a result, asking the tool to create a caricature based on what it knows about me is always going to miss the mark.
I’m not a tech bro.
My desk is immaculate.
Though I do use a lot of data, I don’t smile like that about it.
The tool has just filled in the blanks.
As a prediction model, this is likely because many users of the tool match this output.
It’s why there are examples of amputee users being re-drawn with limbs.
The ‘average’ user has two arms and two legs. So, the model ‘predicts’ that the current individual also meets those parameters and draws that prediction.
It’s regression to the mean in digital form.
And for anyone that understands the origin of regression from the mean, that should be a red flag in of itself. (Google: Francis Galton and eugenics).
Use the tool and have fun.
But also use it as a visual demonstration of its inherent bias and dataset challenges.
