Catherine Heathwood

AI-generated faces are becoming difficult to distinguish from real people, enabling fraud such as a $25 million Hong Kong corporate scam. Research led by Amy Dawel found that adults initially identified AI-generated faces with only 41% accuracy, but a one-hour training session focused on holistic characteristics—including distinctiveness, memorability, expressiveness, symmetry, proportion and attractiveness—raised accuracy to 81%. The method appears promising but has so far been tested mainly with young adults and StyleGAN3-generated white faces, and its effectiveness against newer AI models and across age groups remains uncertain. Human judgment will likely need to evolve alongside automated detection tools as deepfake technology advances.

AI-generated faces are increasingly difficult for people to distinguish from real ones and can be used in scams, including a $25 million deepfake-enabled corporate fraud case in Hong Kong. Research led by Amy Dawel at the Australian National University found that people initially identified AI-generated faces correctly only 41% of the time, but a roughly one-hour training session based on holistic judgments of distinctiveness, memorability, expressiveness, symmetry, proportion and attractiveness raised accuracy to 81%. AI faces often appear unusually average, symmetrical, proportionate, attractive and familiar, while real faces tend to be more distinctive and memorable. The study focused on white faces and young adults and tested StyleGAN3, so its results may not generalize to other models, age groups or demographics. Researchers argue that human training remains important because automated detection tools can fail when AI models change or images are edited, and similar training could eventually help people identify synthetic voices and videos.

Experts describe how human evolution continues through genetic variation, global migration and changing environmental pressures, while modern medicine and social factors reduce traditional selective forces. Increasing reliance on aesthetic procedures shows how appearance can change within a lifetime, while advances in gene-editing technologies like CRISPR raise the possibility of intentionally altering traits in the future. Ethical concerns currently limit genetic modification, but long-term developments could shift views and potentially reshape human evolution, including the possibility of new human lineages in distant environments such as space.

Researchers explain how human physical traits continue to evolve through genetics, environment, and global migration. Changes such as lactose tolerance, skin tone variation, and reduced genetic differences are linked to dietary shifts, environmental conditions, and population mixing. Cosmetic surgery is rising but does not influence genetic inheritance. Scientists note that long-term evolutionary changes are unpredictable and depend on future environmental conditions, with possibilities of divergence if human groups live in isolated environments such as space.