As technology has grown in complexity, it has become hard to tell if images and videos are real or computer generated. These new computer systems, called artificial intelligence (AI), are able to perform tasks that normally require human intelligence, such as mimicking movement, recognizing speech patterns, making decisions, and translating languages.
With society’s newfound ability to easily manipulate images and videos with AI, distinguishing between the real and the artificial has become challenging. This could spell danger if bad actors create false content meant to scare people or spread false information. For example, AI is now able to imitate and digitally alter voices and body movements of prominent figures, such as CEOs and politicians. These videos are called deepfakes, and they can blur the lines of what is real and what is imaginary. Misinformation can be used to sway public opinion during times of crises, or to trick individuals during election season.
That’s why it’s vital to know when images and videos are created by AI. Let’s explore key indicators to help you spot AI and deepfakes, empowering you to navigate the digital landscape more critically.
Inconsistencies in Facial Expressions and Movements
One of the telltale signs of AI-generated content is unnatural facial expressions and movements. Deepfake algorithms struggle to perfectly mimic the nuances of human emotions, resulting in slight glitches or irregularities. In videos, look at blinking patterns, lip synchronization, and facial expressions, as these can be subtly inaccurate. In still images, look at eyes, mouths, hands, and extremities. AI has a hard time depicting natural teeth and eye expressions. It will often add a third arm or extra fingers, and it sometimes twists limbs in odd directions.
Anomalies in Lighting and Shadows
Artificial intelligence struggles with lighting conditions and shadows. Analyze the light sources, shadows, and reflections within the image or video. Inconsistencies such as a shadow on the wrong side of an object could indicate that the image is computer generated.
Unusual Artifacts and Distortions
Artificial intelligence algorithms can create distortions when making images. Keep an eye out for pixelation, blurring, or strange patterns in the image or video. The image may have a line down the center that indicates the image has been expanded with AI, or there may be an odd pattern on part of the image that doesn’t match up with the rest of the pattern.
Lack of Background Consistency
Background elements provide valuable context and reference points for whether the content is real. Deepfakes sometimes warp or distort the subject’s surroundings. Analyze details such as leaves, wallpaper, or curtains to spot any irregularities that might indicate it is AI content.
Examine Audio Cues
While artificial intelligence is mostly used on visual content, deepfakes can have manipulated audio as well. The video could sound remarkably similar to a public figure’s speech patterns, but there are subtle tells that can alert you that the content is fake. Pay attention to any discrepancies in voice tone, pitch, or background noise that may indicate the audio is altered. This video of “Morgan Freeman” is an excellent example of how deepfake technology has become scarily close to reality.
Conduct a Sanity Check
Sometimes, bad actors may want to trick you to believe a public figure has said or done something outrageous, and you have to question the reality of the situation. For example, an image of Bill Gates riding a unicorn is obviously fake and does not indicate that unicorns actually exist.
This is especially important to remember during election season. There is a deepfake of Secretary Hillary Clinton endorsing Governor Ron DeSantis for president, which would be a shocking scenario if it were real – we know that a former Democratic presidential candidate is extremely unlikely to endorse a Republican candidate for the presidency. Regardless of your political affiliation, it is important to know the truth about what candidates are saying and what is made up by opposition groups and individuals.
Employ Technology Solutions
As the technology to create deepfakes advances, so does the technology to detect them. There are many companies creating software designed to detect AI-based tools and deepfakes. Using advanced algorithms, these tools analyze subtle cues and patterns that may not be visible to the human eye. This provides an additional layer of security to help differentiate between authentic and fake content. Currently, deepfake detection tools are geared towards corporate use and are not yet free for public use. There are, however, free AI detection tools for text that can determine if words have been plagiarized.
As AI and deepfake technologies continue to evolve, it becomes crucial for individuals to develop a critical eye on the internet. By understanding the complexities of these technologies and being vigilant in your observations, it is possible to navigate the digital landscape and safeguard ourselves against their potential misuse. Stay informed, stay vigilant, and contribute to the creation of a more discerning digital society.
Other Tech Tips
Want to know more about technology? Read other blogs in our series to learn about technological advancements, internet safety, and how to protect yourself from bad actors online.