Article image

Which face is real? Discerning AI generated images isn't as easy as you think

In the digital age, distinguishing between reality and artificiality has become an unexpected challenge, particularly when it comes to identifying images of real people compared to AI generated images. A recent study by researchers at the University of Waterloo has shed light on just how difficult this task can be.

With the rapid advancement of artificial intelligence (AI), images generated by AI technologies are becoming increasingly difficult to differentiate from photographs of real individuals. This development poses significant questions about our ability to recognize authenticity in the digital realm.

AI generated images and human perception

The study involved 260 participants who were presented with 20 unlabeled images, half of which were photographs of real people obtained from Google searches, and the other half were generated by AI programs such as Stable Diffusion and DALL-E. These programs are renowned for their ability to create highly realistic images.

Participants were tasked with identifying which images were real and which were AI-generated, providing reasons for their decisions.

Surprisingly, only 61% of the participants were able to accurately distinguish between real and AI-generated images, a figure significantly lower than the researchers’ anticipated accuracy rate of 85%.

“People are not as adept at making the distinction as they think they are,” remarked Andreea Pocol, a PhD candidate in Computer Science at the University of Waterloo and the study’s lead author.

Misjudging digital realities

This revelation underscores a growing challenge in the digital age: the increasing difficulty of distinguishing between genuine and artificial content.

Participants in the study focused on details such as fingers, teeth, and eyes as indicators of authenticity. However, their assessments were not always accurate, highlighting the sophistication of AI-generated images.

Pocol pointed out that the study’s context allowed for detailed scrutiny of each photo, a luxury not afforded to the average internet user who typically glances at images briefly. “People who are just doomscrolling or don’t have time won’t pick up on these cues,” Pocol explained.

Evolving battle against disinformation

The rapid pace of AI development further complicates this issue, with the technology advancing faster than academic research and legislation can keep up. Since the study commenced in late 2022, AI-generated images have become even more realistic.

These images pose a particular threat as tools of political and cultural manipulation, enabling the creation of fake images of public figures in potentially damaging scenarios.

“Disinformation isn’t new, but the tools of disinformation have been constantly shifting and evolving,” Pocol observed. She warned of a future where, despite training and awareness, people may still struggle to differentiate between real and fake images.

This potential reality underscores the need for the development of tools to identify and counter AI-generated content. Pocol likened the situation to a new form of AI arms race, emphasizing the importance of staying ahead in the battle against digital deception.

AI generated images and the future of content

In summary, the University of Waterloo’s study shines a spotlight on a crucial challenge in the digital age: our collective struggle to distinguish between real and AI-generated images.

With only a fraction of participants accurately identifying artificial creations, it’s clear that we must enhance our vigilance and develop more sophisticated tools to counteract the rising tide of digital misinformation.

As AI technology continues to evolve at a breakneck pace, staying ahead in this new arms race is imperative for preserving the integrity of our digital world. This calls for a concerted effort from researchers, policymakers, and the public to build a future where we can trust what we see online.

More about AI generated images and disinformation

As discussed above, AI generated images have emerged as a double-edged sword, providing creative opportunities on one hand while ushering in an era of unprecedented disinformation on the other.

Proliferation of AI image creation tools

AI’s ability to create lifelike images has progressed significantly, thanks to advancements in machine learning and neural networks.

Tools like DeepFake, Stable Diffusion, and DALL-E have democratized the creation of highly realistic images, videos, and art that were previously impossible without extensive resources or skills.

While these technologies herald a new age of creativity and efficiency, they also open Pandora’s box of potential misuse, especially in creating deceptive content.

Discerning truth from fiction

The advent of AI generated images has intensified the challenge of disinformation, making it increasingly difficult to discern truth from fabrication. These images can be weaponized to create false narratives, manipulate public opinion, and undermine trust in media and institutions.

The ease and speed with which AI can produce convincing fake content have outpaced the ability of traditional verification methods to keep up, leaving a gap that disinformation can readily exploit.

Real-world implications

The implications of AI generated disinformation are vast and varied, affecting everything from politics to personal reputations. Fabricated images can sway elections, incite violence, defame public figures, and spread conspiracy theories.

The potential for harm escalates as AI technology becomes more accessible and its products more difficult to distinguish from reality. The societal impact is profound, eroding trust and fostering a climate of skepticism and paranoia.

Combating AI generated images and disinformation

Addressing the challenge of AI-generated disinformation requires a multifaceted approach. First, there’s a need for continued development of detection technologies that can keep pace with AI’s advancements. These tools must be integrated into social media platforms and news outlets to identify and flag fake content.

Additionally, public awareness campaigns can educate individuals on the prevalence of AI-generated disinformation and how to critically evaluate the credibility of images and sources. Lastly, policy and regulation need to evolve to address the unique challenges posed by AI, ensuring accountability for creators of malicious content.

The collective challenge ahead

In summary, as we navigate the mirage created by AI-generated images, the battle against disinformation becomes increasingly complex. The potential for these technologies to distort reality and manipulate perceptions underscores the urgent need for robust solutions.

By harnessing advancements in detection technology, raising public awareness, and implementing effective policies, we can mitigate the impact of AI-generated disinformation. The path forward requires collective vigilance and innovation to preserve the integrity of our digital and real-world landscapes.

The full study was published in the journal Advances in Computer Graphics.

Answer: The woman on the left of the photo is NOT real…she’s an AI creation.


Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and


News coming your way
The biggest news about our planet delivered to you each day