Over the last year, or maybe couple of months people got crazy about generative AIs. Most of you probably heard at least once, or read about ChatGPT. This is just one of many generative AIs out there, but it made the front pages the most, so people talk about it a lot. There are many different types of generative AI out there, like ones that create images, music, video, and serve very specific focus and area. I was playing a lot with images last year, when art community exploded around DALL-E, and MidJourney. I even wrote a post about it.
ChatGPT brought a lot of attention, because not only it came in the right time, was used and promoted by the right people, and it aligns with many uses of general public, across different disciplines. People started using it to create copies, research articles, blog posts, social media content and many many more.
Some call it a game changer, and the future. Some are really scared. Others, mainly those who understand LLM (Large Language Models), know that while it looks amazing on the surface, and it has its uses, this is just another logarithmic representation of far fetched dream, and is more like advanced calculator, rather than game changer at all.
I use it at least couple of times a week, to support my pursuit in different projects. ChatGPT is definitely a great tool for research across the web, but only if you keep in mind that it is not perfect. It is actually quite flawed, and unless you accept this fact, and can validate output, then work with it as a draft, probably ChatGPT can do more harm than good. Don’t get scared, it is a very cool engine for many uses, but like library, interviews, stationery, and other things – it is just another tool, and as such shall be treated. Not something that provides a final result, but time saving assistant, that can do many things efficiently and thus make your work easier, more focused on human aspect and context, and allow you to not get distracted by chores, and repetitive tasks, but rather on making your work unique and polished.
Some are scared, because generative AIs become better and better in generating images, videos etc. This is a valid concern, because unless you’re a very detail focused person, you worked in the industry of creating imagery, those AI generated images might fool good chunk of population, if used against them. I happened to work a lot with Photoshop, for different reasons, from restoration of old pictures, to creating photorealistic but fantastic pictures, to doing a digital makeup to enhance photographs, and make them better than they were. My eye can catch a lot of modifications, and so-called improvements, as well as fakes. I can be fooled though as well, and that makes me think, some concerns are valid. If people like me, or even more professional individuals in the industry are wowed by the results of generative AIs, how easy can be an average person manipulated, not knowing tricks, techniques and minute details, that reveal potential manipulation of an image?
There’s no reason to be scared. There’s no reason to be hyped. Even with recent GPT4 version release, ChatGPT is still a nice to have, playful application, useful for light improvements, saving time, and I’d say even better research than most search engines do. This doesn’t mean we shall all not observe this space. There’s no person that shouldn’t be aware of these AIs. They might not replace you, and your job. They might make it a bit easier though, and if they make it easier for someone less skilled, the market will become competitive for cheaper and less skilled workforce, so more advanced individuals should refocus and reposition themselves with their knowledge, to advance in their field, or it might not be enough to have skill and experience, to stay in the same role.
One thing is for sure. I tested couple of ChatGPT equivalents, and ChatGPT itself. I was able to crash it, confuse it, make it generate content on the verge of plagiarism, and content not useful at all. I was able to crash it multiple times, and I wasn’t even trying hard. Other AIs were even less useful. Some repetitive, some complete and utter useless for simple tasks. What’s surprising is that OpenAI struck a deal with Microsoft, and Microsoft implemented some of the functionality of ChatGPT already into their flagship Bing Chat. I must tell you, that one of the simplest tasks I wanted from Bing Chat, ended up a joke, and complete rubbish.
Bottom line, don’t be scared, but be aware. Don’t hype around it, but try to play a bit and find its use to optimise your own work. These AIs are not going anywhere. They are here to stay. They will get better over time. It’s not a question if, but when.