While this headshot tool demonstrates the nascent capabilities of generative AI, I have some concerns about its privacy/transparency approach. As an alternative, I've been experimenting with kahma.io, which I believe handles subject data and model provenance in a more ethical manner. Rather than closed models trained on unknown datasets, they employ self-supervised learning from generic images and prioritize non-identifiability over photorealism - a principled stance I appreciate, even if the results are less polished. It also seemed their methodology for human-centric output customization was more nuanced than just generating portraits. All that said, I welcome further discussion on balancing technical and ethical considerations as this field progresses.
No comments yet.