Meta is reportedly changing its "Made with AI" labels for artificial intelligence-generated photos to a new name, "AI Info," after the labeling system mistakenly identified highly and advanced edited photos as AI-made.
In one well-known instance, former White House photographer Pete Souza revealed to sources that cropping tools enhance the photographs' content, informing Meta's AI detectors.
(Photo : LIONEL BONAVENTURE/AFP via Getty Images)
Meta, on the other hand, stated that it is trying to balance rapidly advancing technology and its obligation to educate users about what its feeds display.
The way Meta approaches things is changing, which highlights how quickly AI technologies are proliferating online and making it harder and harder for the average person to discern what is actually true.
Read Also: Meta CEO Mark Zuckerberg Criticizes Tech Companies Pursuing a God-Like 'One True AI'
People Against AI Media
Notably, a study reported in May found that people are finding it increasingly difficult to distinguish between media provided by AI and media created by humans. This has important ramifications for media literacy and cybersecurity.
Researchers from the CISPA Helmholtz Center for Information Security, in cooperation with other universities, carried out the study, which involved surveying almost 3,000 individuals from Germany, China, and the United States. It is the first comprehensive transnational analysis of this problem.
The study looked into whether people could tell the difference between media produced by AI and actual media. Remarkably, the findings show that people find it difficult to distinguish between the two across various media and demographic contexts.
Participants showed varying ability levels in identifying artificial intelligence (AI)- generated content despite differences in age, education, political convictions, and media literacy.
The study, which took place in three countries between June and September of 2022, used an online survey methodology and exposed participants to real and AI-generated media samples.
Despite attempts to collect a wide range of sociobiographical information and evaluate variables like political inclination and media literacy, most participants mislabeled AI-generated media as human-made.
Realistic AI Media
The study emphasizes how frighteningly adept AI is at producing realistic texts, photos, and audio recordings. Dr. Lea Schönherr, Professor Dr. Thorsten Holz, and his colleagues clarified the dangers of AI-generated content. In particular, they discussed how the material could be abused to sway public opinion and control political dialogue.
The ease with which large volumes of media content may now be created because of the quick development of artificial intelligence raises questions about its misuse, particularly during important events like elections.
The many elections this year, especially the forthcoming U.S. presidential elections, make this worry much more acute.
To lessen the threats to democracy, Dr. Thorsten Holz highlights the urgent necessity for automatic recognition of AI-generated media. Dr. Schönherr does, however, draw attention to the difficulty presented by the developing AI generation techniques, which make it more challenging to recognize such information automatically.
Related Article: Microsoft's AI Chief Sparks Controversy with 'Freeware' Claims
