Five years ago, if one were to ask someone when they thought humankind would develop a functional AI model, they would say that such technology was decades, perhaps even centuries away. The truth is, although AI is still in its very crude stages, it can still perform feats today’s society never expected to witness in their lifetime. Hence, it should come as no surprise that humans are woefully unprepared to deal with this new, stellar technological leap, with practically no regulations to check its usage.
If AI were limited to natural language generation, it would still be relatively easy to control. However, AI can already generate images based on natural language descriptions. How it manages to achieve this is where the controversy starts: these models have to comb through hundreds of millions of photographs and artworks to learn to how identify objects, patterns, and styles. This means that to train a model, images of thousands of artists need to be scanned, often without their consent. And at the end of the day, these models mimic the said artists in a way that is almost indistinguishable from the original. This means that not only are the intellectual property rights of these artists violated, but they also stand the risk of losing their jobs.
But this is only one way of looking at it. AI image generators make art accessible to everyone, even those who could not normally afford such a luxury. It also allows individuals to create the art pieces they envision exactly the way they picture them, without having to spend a decade learning the necessary skills. Furthermore, collectors who pay hundreds, even thousands of dollars on paintings are unlikely to settle for a framed printout instead of actual paint and canvas. Thus, artists will not lose their primary customers - people who understand and appreciate art and can afford it.
If we are to progress as a race, we must embrace new technologies as they develop, while striking a delicate balance between liberty and regulation. The real threat is not how generative AI can create stunning artworks: it is how this technology, in the wrong hands, can be used for defamation and hate-mongering. It can be used to lend credibility to fake news by rendering almost life-like illustrations of fictitious scenes.
Hence, two measures need to be enacted to check the growing power of these AI models. Firstly, artists should have the right to opt out of having their artworks used for training data sets. This will help protect their intellectual property rights. Secondly, tools should be developed that can be used to distinguish AI-generated images from regular ones. This can go a long way in preventing the usage of AI for defamation and spreading misinformation.
Of course, this is not a permanent solution. Every month, AI sees new developments and implementations. Policymakers must acknowledge the growing importance of curbing the power of this tool, without having to resort to banning it like Italy did to ChatGPT. Therefore, the world must move forward with an open mind, but with the necessary regulations in place.