WHICH HUMAN FACES CAN AN AI GENERATE? LACK OF DIVERSITY IN THIS PERSON DOES NOT EXIST
Bibliographic record
Abstract
In this abstract we show the results of an interdisciplinary research in which we audit fake human faces generated by the website This Person Does Not Exist (TPDNE), and discuss how this system can help perpetuate normativities supported by a dependency on a limited database. Our analysis is centered on the “default generic face” that we created by overlapping random samples of fake human faces generated by TPDNE's algorithms – a version of Generative Adversarial Network, the StyleGAN2. To carry these experiments, we built a database with 4100 fake human faces taken from TPDNE via web scraping; we analysed them through a Python language script; and discussed behaviours identified in results. Our analyses are based on the use of images, called “cluster-images”, created from this overlapping of N arbitrary fake human faces by the TPDNE's algorithm. Our experiments showed that, independently of the group of fake human faces sampled, the same generic white face always appeared as a result. These results intrigue particularly because the lack of diversity of TPDNE's generated faces is not a mere problem to be fixed in this system in this digital infrastructure, but a dynamic of reinforcing standards that historically regulate bodies, territories and practices.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.013 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.006 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".