Diffusion Models as Artists: Are we Closing the Gap between Humans and Machines?

Victor Boutin, Thomas Fel, Lakshya Singhal, Rishav Mukherji, Akash Nagaraj, Julien Colin, Thomas Serre
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:2953-3002, 2023.

Abstract

An important milestone for AI is the development of algorithms that can produce drawings that are indistinguishable from those of humans. Here, we adapt the ”diversity vs. recognizability” scoring framework from Boutin et al (2022) and find that one-shot diffusion models have indeed started to close the gap between humans and machines. However, using a finer-grained measure of the originality of individual samples, we show that strengthening the guidance of diffusion models helps improve the humanness of their drawings, but they still fall short of approximating the originality and recognizability of human drawings. Comparing human category diagnostic features, collected through an online psychophysics experiment, against those derived from diffusion models reveals that humans rely on fewer and more localized features. Overall, our study suggests that diffusion models have significantly helped improve the quality of machine-generated drawings; however, a gap between humans and machines remains – in part explainable by discrepancies in visual strategies.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-boutin23a, title = {Diffusion Models as Artists: Are we Closing the Gap between Humans and Machines?}, author = {Boutin, Victor and Fel, Thomas and Singhal, Lakshya and Mukherji, Rishav and Nagaraj, Akash and Colin, Julien and Serre, Thomas}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {2953--3002}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/boutin23a/boutin23a.pdf}, url = {https://proceedings.mlr.press/v202/boutin23a.html}, abstract = {An important milestone for AI is the development of algorithms that can produce drawings that are indistinguishable from those of humans. Here, we adapt the ”diversity vs. recognizability” scoring framework from Boutin et al (2022) and find that one-shot diffusion models have indeed started to close the gap between humans and machines. However, using a finer-grained measure of the originality of individual samples, we show that strengthening the guidance of diffusion models helps improve the humanness of their drawings, but they still fall short of approximating the originality and recognizability of human drawings. Comparing human category diagnostic features, collected through an online psychophysics experiment, against those derived from diffusion models reveals that humans rely on fewer and more localized features. Overall, our study suggests that diffusion models have significantly helped improve the quality of machine-generated drawings; however, a gap between humans and machines remains – in part explainable by discrepancies in visual strategies.} }
Endnote
%0 Conference Paper %T Diffusion Models as Artists: Are we Closing the Gap between Humans and Machines? %A Victor Boutin %A Thomas Fel %A Lakshya Singhal %A Rishav Mukherji %A Akash Nagaraj %A Julien Colin %A Thomas Serre %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-boutin23a %I PMLR %P 2953--3002 %U https://proceedings.mlr.press/v202/boutin23a.html %V 202 %X An important milestone for AI is the development of algorithms that can produce drawings that are indistinguishable from those of humans. Here, we adapt the ”diversity vs. recognizability” scoring framework from Boutin et al (2022) and find that one-shot diffusion models have indeed started to close the gap between humans and machines. However, using a finer-grained measure of the originality of individual samples, we show that strengthening the guidance of diffusion models helps improve the humanness of their drawings, but they still fall short of approximating the originality and recognizability of human drawings. Comparing human category diagnostic features, collected through an online psychophysics experiment, against those derived from diffusion models reveals that humans rely on fewer and more localized features. Overall, our study suggests that diffusion models have significantly helped improve the quality of machine-generated drawings; however, a gap between humans and machines remains – in part explainable by discrepancies in visual strategies.
APA
Boutin, V., Fel, T., Singhal, L., Mukherji, R., Nagaraj, A., Colin, J. & Serre, T.. (2023). Diffusion Models as Artists: Are we Closing the Gap between Humans and Machines?. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:2953-3002 Available from https://proceedings.mlr.press/v202/boutin23a.html.

Related Material