When talking about the dangers of AI, its use in disinformation tasks is always mentioned, and with good reason. Beyond dystopian speculations like Skynet, a very present reality is that with it Different types of content can be displayed in a more than convincing way with the necessary tools and skills. To sum it all up in one term, they are known as deepfakes.
Microsoftwhich defines a deepfake as an AI-generated image, video, or audio recording used to spread false information, has launched a website that tests visitors’ ability to distinguish between AI-created images and real ones. It was presented this week by its president, Brad Smithit’s free and it’s called Real or Not?
Its operation is very simple. As you access the Weba brief explanation states that ‘the increasing quality of AI images makes them harder to detect. Can you tell if this image is real or AI-generated?’
After clicking on Start GameReal or Not presents the visitor with a succession of 15 images and in each of them you have to say if it is real or artificial by clicking on the corresponding button. After each choice, the website indicates whether you have guessed correctly or not and at the end of the series it provides the total percentage of correct answers and compares it with the average of the visitors. If the user wants to make another attempt, they will find with a new image gallery different from the previous one.
Generative artificial intelligence is still a recent phenomenon that is taking its first steps, but in the same way that it has advanced extraordinarily in the last three years, it will continue to do so and making it even more difficult for Internet users in the coming years.
There are factors that help distinguish them, such as the fact that they often result too polishedsuch as images treated with Photoshop in the 90s, or elements that are not represented correctly as texts or the famous problem of the fingers of the hands that in 2024 is already something residual, but a couple of years ago it was common. In any case, As the months go by, they are becoming better and more deceptive about their nature..
Microsoft has also created an educational resources page with information and tips to help people better deal with deepfakes, which are likely to be very prevalent in the coming weeks in light of the US presidential election. Unfortunately, There is no foolproof trick that helps to distinguish them, but you have to rely on common sense.These tips include checking sources, verifying information before sharing it, reporting deepfakes you find, and staying up to date on the development of these technologies and what they can do.