Recent Research Unveils 35% of Americans Cannot Differentiate Between Real Pictures and Artificial Intelligence

As AI generation software improves, it becomes more difficult to distinguish between what is genuine and what is not.

03:15 16 March 2023

As AI generation software improves, it becomes more difficult to distinguish between what is genuine and what is not. Is it possible for the general public to distinguish between authentic and manufactured, renowned landmarks?

Concured, a cutting-edge artificial intelligence-powered content recommendation system, questioned 1,000 Americans to discover if AI is capable of producing realistic-looking images. What would be a better way to test it than by employing popular attractions that we all have become accustomed to?

The development of algorithms has significantly increased the ability of artificial intelligence (AI) and machine learning to produce realistic images, making it much more challenging to distinguish between AI-generated images and those taken by actual cameras. Generative adversarial networks (GANs) are a class of algorithms that have been used to produce images of everything from people and animals to objects and landscapes. 

The ability of GANs to learn from vast quantities of actual images is one reason it is challenging to distinguish between real and AI-generated images. These datasets may contain millions of images, giving the algorithm a wealth of knowledge about how genuine photos appear. As a result, GANs can create pictures that are not only visually similar to real pictures but also capture the complexity and underlying structure of sincere pictures, making them even more difficult to tell apart from legitimate pictures.

Additionally, with each new GAN generation, the quality of AI-generated images gets better. It is getting harder and harder to separate artificial images from real ones as algorithms become more sophisticated and potent and better able to capture the nuances and details of real images.

 

The Concured poll, which discovered that nearly a third of people couldn't tell the Eiffel Tower apart from its real counterpart and that more than a third of Americans thought all of the AI-generated images were their real counterparts, supports this.

“AI will change marketing and content. It already automates complex tasks, personalises interactions, and offers data-driven insights, resulting in increased effectiveness, profitability, and efficiency. 

It already highlights the power of AI and the difficulty of detecting AI-created content when 35% of Americans are unable to distinguish between real and AI images”, said Tom Salvat, Concured's founder and CEO.

One of the main advantages of AI-generated images is their speedy production of a large number of distinctive images, which can be useful in industries like advertising and e-commerce. 

It is important to keep in mind that AI-generated images still have issues and differ from real images. AI-generated images could sometimes have less detail and appear a little bit pixelated or distorted than truthful photographs. 

 

Answers

Left: AI 

Right: Real