Generative AI-based image editing tools from Google, Adobe are altering reality one feature at a time

The year 2023 has witnessed a sudden surge in the use of generative AI-based apps and applications. From content creation and analysing large swathes of data for predicting future trends to generating images and videos, generative AI today is being deployed across industries for a variety of purposes. While industries are still finding new use-cases for generative AI, one use-case that has sounded alarm bells among the global tech community is the image editing features that this technology has fostered.

Image editing software and tools have been available in the market for decades now. While some like Microsoft Paint, Fotor, and PicMonkey let users add effects and text, blur background, adjust colour-based properties and make collages among others, others like Adobe Photoshop and Pixlr give users more freedom in modifying backgrounds and even using elements from one image in another. But now, generative AI has paved the way for a new crop of image editing features that can not only modify images as traditional tools have been doing so far, albeit more easily, but also alter reality as we know it.

Enter Google Pixel 8 series.

Google recently unveiled its Pixel 8 series, which includes the Pixel 8 Pro and the Pixel 8 smartphones. While the Pixel smartphones have always been known for their camera prowess thanks to the company’s powerful AI-based algorithms, this year Google took the stakes much higher by adding generative AI to the mix.

This year, the company introduced three new generative AI-based features exclusive to the Pixel 8 series, that can edit expressions, voice and reality, as was captured in an image. Here’s a quick brief about these features:

– Best Take feature in Google Photos uses a series of similar photos taken close together to help users automatically create a blended image with everyone’s best expression. Users can also manually select another expression from the clicked images.

– Magic Editor uses generative AI to help users make complex edits such as repositioning the subject, changing background colouring or sky colour. It can also remove objects from an image to generate an image that is closer to their vision of how they want the image to look like. Here’s an example:

Image: Google

– Audio Magic Eraser uses advanced machine learning models to identify sounds, like people talking in the background or wind, to sort them into layers that can be controlled or even reduced if needed.

That said, Google is not the only company that has come with some outrageous image-editing features.

Enter Adobe and its army of new projects.

The company recently hosted its Adobe Max conference in Los Angeles wherein the company introduced a host of new generative AI-based projects and tools that take image and voice editing features offered by any tool to a whole new level. Here is a quick brief about these projects:

– Project Stardust enables users to easily select, edit and even delete complex elements in any image. It lets users select persons in a photograph, move them to a different place in the composition and fill in the background where they were previously standing. Users can also change elements like the colour of a person’s clothing or the position in which they are standing. Here’s an example of its capabilities:

Notice how the tool recreates the part of the image that would otherwise have been missing?

– Project Dub Dub Dub automates the video dubbing process. It uses AI capabilities, a recording or audio track of a video that can be automatically translated to all supported languages while preserving the voice of the original speaker, temporally aligned with the original dialogue and ready to publish.

– Project Scene Change enables users to composite a subject and scene, from two separate videos captured with different camera trajectories, into a scene with synchronised camera motion.

Additionally, Adobe last month announced new Firefly web applications that use generative AI models for adding images, text effects and vectors via text-based prompts in over 100 languages.

There’s an argument to be made against such tools.

Sure, these features do have some benefits. For instance, Pixel 8 series’ Best Take feature lets users capture that one image that is often necessary yet incredibly difficult to capture, especially when there are a lot of people in a photo. It also lets users sync their expressions for a moment (staged or not) that users want to capture. Similarly, the Audio Magic Eraser tool lets users crop out the unwanted noise so that the message in their video is clearer. Also, Adobe’s Project Dub Dub Dub takes away the hefty cost of dubbing a video in multiple languages. Its Project Stardust, on the other hand, enables users to draft a moment as they would have wanted it to be. But there is an argument to be made against the use of such features.

These tools not only enable users to edit the images and videos that they captured, but they also let them craft a version of reality that suits their preferences or in other words create an alternate reality that never existed in the first place. This might seem harmless at first, but in the longer run it will make it difficult for people to distinguish between what’s real and what’s not. And in the hands of wrong people, it can also have negative or harmful consequences.

For instance, these tools can be used to efficiently crop out a person or an object from an image, which in turn can tamper with justice. In another scenario, these tools can be used for spreading fake news more effectively — an issue that tech giants such as X (formerly Twitter) and Meta have been grappling with for a long time now. Similarly, these tools can also be used for cyber-warfare by criminals. Deepfakes remain one of the more concerning issues of our times and are often used by cyber-criminals to alter reality, malign images of public personalities and spread fake news. They are also used by countries for cyber-espionage.

Making these tools available easily will only make it easier for cyber-criminals to spread fake news and tamper with evidence interfering with justice. It will also make it that much more difficult to distinguish between what’s real and what’s not.

While a lot of these tools aren’t still available to the wider public and in some cases, they require users to have specific devices in hand, their availability, even in the near future, poses a threat to reality and to the society at large.