AI faces look more real than actual human faces

AI faces look more real than actual human faces

White faces generated by artificial intelligence (AI) now appear more real than human faces, according to new research from The Australian National University (ANU) and the University of Aberdeen. 

In the study, more people thought AI-generated White faces were human than the faces of real people. The same wasn’t true for images of people of colour. 

The reason for the discrepancy is that AI algorithms are trained disproportionately on White faces, Dr Amy Dawel, the senior author of the paper, said. 

“If White AI faces are consistently perceived as more realistic, this technology could have serious implications for people of colour by ultimately reinforcing racial biases online,” Dr Dawel said. 

 “This problem is already apparent in current AI technologies that are being used to create professional-looking headshots. When used for people of colour, the AI is altering their skin and eye colour to those of White people.” 

 One of the issues with AI ‘hyper-realism’ is that people often don’t realise they’re being fooled, the researchers found. 

 “Concerningly, people who thought that the AI faces were real most often were paradoxically the most confident their judgements were correct,” Elizabeth Miller, study co-author and PhD candidate at ANU, said. 

 “This means people who are mistaking AI imposters for real people don’t know they are being tricked.” 

 The researchers also discovered shy AI faces are fooling people. 

 “It turns out that there are still physical differences between AI and human faces, but people tend to misinterpret them. For example, White AI faces tend to be more in-proportion and people mistake this as a sign of humanness,” Dr Dawel said. 

 “However, we can’t rely on these physical cues for long. AI technology is advancing so quickly that the differences between AI and human faces will probably disappear soon.” 

 The researchers argue this trend could have serious implications for the proliferation of misinformation and identity theft, and that action needs to be taken. 

 “AI technology can’t become sectioned off so only tech companies know what’s going on behind the scenes. There needs to be greater transparency around AI so researchers and civil society can identify issues before they become a major problem,” Dr Dawel said. 

 Raising public awareness can also play a significant role in reducing the risks posed by the technology, the researchers argue. 

 “Given that humans can no longer detect AI faces, society needs tools that can accurately identify AI imposters,” Dr Dawel said. 

 “Educating people about the perceived realism of AI faces could help make the public appropriately sceptical about the images they’re seeing online.” 

Dr Clare Sutherland from the University of Aberdeen who co-authored the paper added  “As the world changes extremely rapidly with the introduction of AI, it’s critical that we make sure that no one is left behind or disadvantaged in any way - whether due to their ethnicity, gender, age, or any other protected characteristic.” 

The study was published in Psychological Science, a journal of the Association for Psychological Science 

 

ENDS 

 

Notes to Editors 

Search News

Browse by Month

2024

  1. Jan
  2. Feb There are no items to show for February 2024
  3. Mar
  4. Apr
  5. May
  6. Jun There are no items to show for June 2024
  7. Jul There are no items to show for July 2024
  8. Aug There are no items to show for August 2024
  9. Sep There are no items to show for September 2024
  10. Oct There are no items to show for October 2024
  11. Nov There are no items to show for November 2024
  12. Dec There are no items to show for December 2024

2023

  1. Jan There are no items to show for January 2023
  2. Feb
  3. Mar There are no items to show for March 2023
  4. Apr
  5. May
  6. Jun
  7. Jul There are no items to show for July 2023
  8. Aug
  9. Sep There are no items to show for September 2023
  10. Oct There are no items to show for October 2023
  11. Nov
  12. Dec

2022

  1. Jan There are no items to show for January 2022
  2. Feb There are no items to show for February 2022
  3. Mar There are no items to show for March 2022
  4. Apr There are no items to show for April 2022
  5. May There are no items to show for May 2022
  6. Jun There are no items to show for June 2022
  7. Jul There are no items to show for July 2022
  8. Aug
  9. Sep
  10. Oct
  11. Nov There are no items to show for November 2022
  12. Dec There are no items to show for December 2022

2020

  1. Jan There are no items to show for January 2020
  2. Feb There are no items to show for February 2020
  3. Mar
  4. Apr There are no items to show for April 2020
  5. May There are no items to show for May 2020
  6. Jun There are no items to show for June 2020
  7. Jul There are no items to show for July 2020
  8. Aug There are no items to show for August 2020
  9. Sep There are no items to show for September 2020
  10. Oct There are no items to show for October 2020
  11. Nov There are no items to show for November 2020
  12. Dec There are no items to show for December 2020

2019

  1. Jan There are no items to show for January 2019
  2. Feb There are no items to show for February 2019
  3. Mar There are no items to show for March 2019
  4. Apr There are no items to show for April 2019
  5. May There are no items to show for May 2019
  6. Jun There are no items to show for June 2019
  7. Jul
  8. Aug
  9. Sep There are no items to show for September 2019
  10. Oct
  11. Nov There are no items to show for November 2019
  12. Dec There are no items to show for December 2019

2018

  1. Jan
  2. Feb
  3. Mar
  4. Apr
  5. May There are no items to show for May 2018
  6. Jun
  7. Jul There are no items to show for July 2018
  8. Aug There are no items to show for August 2018
  9. Sep There are no items to show for September 2018
  10. Oct There are no items to show for October 2018
  11. Nov There are no items to show for November 2018
  12. Dec There are no items to show for December 2018

2017

  1. Jan There are no items to show for January 2017
  2. Feb
  3. Mar There are no items to show for March 2017
  4. Apr There are no items to show for April 2017
  5. May There are no items to show for May 2017
  6. Jun There are no items to show for June 2017
  7. Jul There are no items to show for July 2017
  8. Aug There are no items to show for August 2017
  9. Sep
  10. Oct There are no items to show for October 2017
  11. Nov There are no items to show for November 2017
  12. Dec There are no items to show for December 2017

2016

  1. Jan
  2. Feb
  3. Mar
  4. Apr There are no items to show for April 2016
  5. May
  6. Jun
  7. Jul
  8. Aug
  9. Sep
  10. Oct
  11. Nov
  12. Dec There are no items to show for December 2016

2013

  1. Jan
  2. Feb There are no items to show for February 2013
  3. Mar
  4. Apr
  5. May
  6. Jun There are no items to show for June 2013
  7. Jul
  8. Aug
  9. Sep
  10. Oct
  11. Nov
  12. Dec

2012

  1. Jan There are no items to show for January 2012
  2. Feb There are no items to show for February 2012
  3. Mar
  4. Apr
  5. May
  6. Jun There are no items to show for June 2012
  7. Jul There are no items to show for July 2012
  8. Aug
  9. Sep There are no items to show for September 2012
  10. Oct
  11. Nov
  12. Dec There are no items to show for December 2012

2011

  1. Jan
  2. Feb
  3. Mar
  4. Apr There are no items to show for April 2011
  5. May
  6. Jun
  7. Jul There are no items to show for July 2011
  8. Aug
  9. Sep
  10. Oct There are no items to show for October 2011
  11. Nov There are no items to show for November 2011
  12. Dec There are no items to show for December 2011