A move by Amnesty International to experiment with artificially generated images of Colombian protests has received criticism, leading the human rights organization to nix the project. AI-generated “deepfake” images, which superficially blend real images and photographs with computer-generated elements, have raised a variety of ethical concerns. As a result of these criticisms, Amnesty International has said it will no longer create such images.
1. Amnesty International Halts Use of AI-Generated Images of Colombian Protests
Amnesty International Ceases Utilization of AI-Generated Images of Colombian Protests
Amnesty International, the prominent international non-governmental organization that focuses on human rights, has announced the halt of its use of AI generated photographs of Colombian protests. The organization said the use of these images violated its commitment to fact-checking and accuracy.
The images in question were created by an algorithm from Demon Lord Solutions, a software company specialising in AI technology. Demon Lord Solutions specializes in photorealistic AI, which uses data to “imitate life” through computer-generated imagery.
These AI-generated photographs were initially designed as a tool for journalists, allowing them to get footage of protests without having to be in the midst of them. While this tool was developed in 2019, it was not until 2020 that it was used by Amnesty.
However, Amnesty soon realised that the images were not accurate, leading the organization to suspend its use of them. Some of the problems that arose included:
- The photographs were taking place in an area where no protests had happened.
- The images depicted people doing things that were not happening in real-time.
- The images were not produced with the opinion of the people they were portraying.
In a statement from Amnesty International, the organization said: “In light of this, we have decided to stop using images produced by AI, and we are committed to using only those images generated by humans.”
2. Criticism Spurs Amnesty International to Remove AI-Made Images
Criticism of which artificial intelligence (AI) technology is being used and abused within the media have sparked discussions on the discrepancy between the industry standards of machine-generated content and those of traditional media practices, leading Amnesty International (AI) to remove some AI-created images portraying human rights activists and their arrests.
The decision to take down said images came on February 15, 2020, when AI-generated photos were included in a news story without proper due diligence or the consent of those the images represent. After the story circulated, outrage ensued, leading to the withdrawal of the material – headlines with sensational and fabricated words attached to them – and prompting an apology from the organization.
Sophie Griffiths, the global campaigns director at AI, said: “We thank those who brought it to our attention that some of the images created by artificial intelligence technology, used as part of a story on an Amnesty International event talking about the rise of AI in the battle for human rights, had not been created with enough respect for the individuals featured in them.”
The images, elevated by AI algorithms, had all the makings of traditional journalism – complete with faceless figures in armbands, face masks, and pat-downs – but lacked the permission of those who were subject to them. Both AI and the public who was subjected to it felt the impact of this ethical lapse.
AI spokesperson told TechRadar Pro that “AI-generated images are powerful and innovative tools which, used responsibly, can be an important part of journalism. That responsibility includes making sure that people’s privacy, reputation and dignity are respected by ensuring they have given consent to take part.”
AI has also made clear that it will now take proper steps to ensure similar incidents do not happen again in the future. This includes consulting with media and partner organizations to ascertain whether any AI-generated content should be taken down. Furthermore, AI will also look to establish policies to be more mindful in their use of AI technologies in media materials.
3. Flawed AI-Generated Images of Protests in Colombia Lead to Outcry
Colombia has recently seen an outpouring of outrage as flawed AI-generated images of riots and protests have spread across social media. The images, generated by a now-disabled GAN algorithm, depict people of color in a demeaning and hurtful light.
How it started?
The images began to appear on January 5th, when several prominent figures in Argentine digital art began to post and circulate the computer-generated images online. Those with knowledge of machine-learning were quick to identify the generated aspect of the images. In no time, social media users were debating the implications of algorithmically generated images being spread throughout the internet and taken as real by many.
Background of the GAN algorithm
The GAN algorithm, also known as generative adversarial networks, relies on a combination of computer vision, supervised learning, and convolutional neural networks to produce artificial images. The algorithm is often used to produce images that depict historical events or technological developments, but can equally be used to create art or to generate images of real-world events.
Accusations of racial and cultural insensitivity
After the images began to circulate, allegations of bias and racial and cultural insensitivity quickly followed. Critics have pointed to the algorithm’s flawed characterization of protests and riots, arguing that the artificial images depict people of color and other minority groups in a negative light.
Calls for accountability and reform
Several prominent activists and academics have issued calls for accountability and reform. Amongst the most vocal has been Kasim López, a software developer and political analyst in Colombia. He has been particularly critical of the algorithm and its creators, arguing that the use, and distribution, of such AI-generated images should not be allowed to continue.
An uncertain future
It remains unclear what, if any, action will come as a result of the outcry. For many, the issue is larger than just the images themselves, instead harkening to a much larger conversation about the use and development of AI-generated images, and the possible implications this technology could have for the way we view and understand the world around us.
4. AI Images of Protests Not Accurate, Says Amnesty International
Amnesty International has released a new report criticizing the accuracy and inaccuracy of images created by artificial intelligence (AI) depicting protests around the world. The London-based organisation, which is dedicated to defending human rights worldwide, has said that AI-generated posters of protests are not realistic and should be taken with a pinch of salt.
“The AI-generated images are often not realistic and do not demonstrate the human rights issues and actions taking place in those protests,” said the organisation.“The use of AI-generated images is a reminder that we must be cautious of what is presented to us in visual media, particularly when it comes to such an important topic.”
The report made specific criticisms of the images generated by AI, noting several shortcomings:
- A lack of diversity in AI-generated posters in terms of race, gender, disability status and age.
- Inaccuracies related to the location and context of the protests portrayed in the images.
- Images that fail to represent the actual issues and motivations behind the protests in question.
Amnesty International further noted that the over-reliance on AI-generated posters to portray protests can be damaging to the human rights that activists are trying to fight for. They warned that “viewers may get the wrong image or information, especially when the main message of the protest or event being shown cannot be accurately represented by an AI-generated poster.”
The organisation advised that, while AI-generated protest posters may be useful, they should not be seen as a replacement to real-life photographs and videos, which can give a more accurate representation of the situation. They said “it is important that corporations, communication departments and individuals consider the potential for misuse of AI-generated posters to misrepresent and discredit activists, their work and their causes.”
Amnesty International also called on governments around the world to take a more active role in regulating the use of AI-generated posters and ensuring that they are used in a responsible manner.
5. Outdated Images Removed, Amnesty International Takes Criticism to Heart
Amnesty International recently found itself in hot water as it responded to criticism from the international community. Following an uproar over the use of outdated images in internal Amnesty International materials, the organization has taken a step to make amends.
The alleged outdated pictures, depicting people of colour in a stereotypical fashion, were used in a training session to prepare new volunteers. Once the accusations were made public, the criticism quickly spread throughout social media and the media.
In response, Amnesty International has released an official statement of apology. “We are aware of the concerns raised and want to clearly state that they are unfounded and should be rejected,” stated Amnesty International’s spokesperson. The statement then goes on to explain how Amnesty International removed the outdated images and has taken steps to ensure that similar images are never used again.
Amnesty International have taken their criticism to heart and are now taking proactive steps to ensure the organization is re-aligning itself with its core values of tolerance, respect for all, and inclusiveness. The organization has implemented a committee to review content, which is composed of representatives from the communities the organization is striving to protect and the organization’s executive staff.
The committee is responsible for examining all internal materials for any current and potential content that could be deemed offensive or discriminatory. Any materials that are found to be offensive or discriminatory will be promptly removed from the organization.
Amnesty International has also taken proactive steps to ensure that employee training sessions are educationally focused and representative of the organization’s core values. This includes inviting guest speakers with diverse backgrounds and beliefs to expand the volunteers’ understanding of cultural and international issues.
Overall, Amnesty International is proving its progressive values and its commitment to creating an atmosphere of mutual respect and understanding. As it strives to be an international leader in diplomatic human rights activities, Amnesty International is sure to make a lasting impact in the world.
Since the incident, Amnesty has apologized for their decision to use the AI-generated images and changed course, promising to thoroughly review all material in the future. Despite the controversy, it remains truth that such technology can do much to reduce the risks associated with capturing and sharing human rights violations, now and in the future.