Google Tests Tool to Spot AI Images

Tech giant Google is currently testing a tool to aid in spotting Artificial Intelligence (AI) images, to reduce the spread of fake images.

According to Google, SynthID will help to resolve the issue of using AI image generators to create ‘deepfakes’ false images that mislead the public.

Further, the new tool will add a digital “watermark” to the image that can’t be seen by the human eye but can be picked up by a computer that has been trained to read it.

The firm revealed that the new watermarking tech is resistant to tampering, making it a key step toward policing the spread of fake images and slowing the dissemination of disinformation.

Google introduces SynthID, a tool to spot AI images, — Pushmeet Kohli, vice president of research at Google DeepMind. PHOTO/Courtesy.

“The goal is to help create a system where most AI-created images can be easily identified using embedded watermarks,” said Pushmeet Kohli, vice president of research at Google DeepMind, the company’s AI lab.

Kohli, further cautioned that the new tool is not dependable.

Watermarking is one of the ideas that tech companies are rallying around as a potential way to decrease the negative impact of the “generative” AI.

Consequently, Microsoft has started a coalition of tech and media companies to develop a common standard for watermarking AI images.

They have further disclosed that they are researching new methods to track AI images.

The company also places a small visible watermark in the corner of images generated by its AI tools. OpenAI, whose Dall-E image generator helped kick off the wave of interest in AI last year, also adds a visible watermark.

AI researchers have suggested ways of embedding digital watermarks that the human eye can’t see but can be identified by a computer.

According to Kohli, Google’s new tool is better because it works even after the image has been significantly changed.

How Google SynthID Will Work

SynthID will provide a digital watermark that is woven directly into the pixel data.

Moreover, the tool will incorporate two AI models trained together, one tasked with adding imperceptible watermarks and another for identifying them.

Google optimized these models to embed watermarks that align with the original image content, maintaining visual quality while enabling detection.

Also Read: Kyrgyzstan Bans TikTok Citing Risk to Children

What is AI

According to research by Investopedia, Artificial intelligence (AI) refers to the stimulation of human intelligence by software-coded heuristics.

In 2022 AI emerged in the mainstream through widespread familiarity with applications of Generative Pre-Training Transformer.

The most popular application is OpenAI’s ChatGPT, its widespread fascination made it synonymous with AI in the minds of most consumers.

Also Read: Met Department Warns of El Nino Rains

Types of AI

AI is available in four types.

1.Reactive AI uses algorithms to optimize outputs based on a set of inputs. Chess-playing AIs, for example, are reactive systems that optimize the best strategy to win the game.

2.Limited memory AI can adapt to experience or update itself based on new observations or data.

3.Theory-of-mind AI are fully adaptive and have an extensive ability to learn and retain past experiences.

4.Self-aware AI, as the name suggests, become sentient and aware of their own existence.