Apple has once again demonstrated its ability to deliver exactly what the market needs, and likely in the best way, with its new suite of artificial intelligence tools known as "Apple Intelligence." One of the most notable features of this suite, which will arrive with iOS 18, is the Image Playground application. This app will allow us to generate images from simple text commands or even based on a photo, but with an interesting twist: all generated images will be illustrations. This strategic decision by Apple addresses several crucial points regarding AI risks.
A strategy more interesting than it appears at first glance
Firstly, the decision not to allow the creation of photorealistic images is an effective measure to prevent the spread of false information on the internet. At a time when images can be easily manipulated, the credibility of an image is key to deceiving or misleading. By making all image generation by Apple Intelligence in the form of illustrations, drawings, or animations, these concerns are removed from the equation.
Images can still be used for satire, even political satire, and can be used to illustrate articles or social media posts without losing their utility. At the same time, they avoid the risks associated with photorealism. In this vein, Apple plans to use metadata on the generated images so that anyone can know they were created by Apple Intelligence. This is an additional layer of transparency and security that, although it can be bypassed by simply taking a screenshot, is again limited by the realism of certain uses.
Another important aspect is that with the current state of AI, generating illustrations is significantly easier than producing realistic images. While AI models are powerful, they are not perfect and can make mistakes that are very evident in a photorealistic image, but much more acceptable, if not invisible, in an illustration. The style itself allows for greater flexibility and tolerance for inaccuracies. Issues such as the size of the eyes or an unnatural smile do not attract attention in an illustration, with the greatest example of this being caricatures.
With these decisions, Apple covers two very important fronts: offering a satisfactory result in many more instances and reducing the inherent risks of AI. By avoiding the creation of photorealistic images, Apple ensures that its tool cannot be used for certain malicious purposes, which would undoubtedly generate headlines quickly. On the other hand, while it is expected that models will continue to improve and photorealism will get better, in the meantime, we have truly impressive images that are likely to enliven a conversation in the Messages app or a social media post with ease and without having to try too many times.
Overall, Apple has been very smart with its approach to image generation through Apple Intelligence. An approach that demonstrates a clear understanding of the challenges and responsibilities involved in implementing AI tools in everyday devices like iPhones. An implementation that allows us to engage with technology, without fear, in a way that, when used well, is truly interesting and very fun.
Ot Hanaringo | How to install iOS 18 on iPhone: developer beta