You’ve seen the Skynet memes for a few years now, and with every advancement in what the mainstream calls artificial intelligence (AI), the more the fears of the tech robbing people of jobs crop up. While the fear is not an irrational one, it’s certainly not happening anytime soon. And recently, another example of just this has appeared in the news.
Last week, Game Developer reported that Irish video games company Keywords Studios shared in its fiscal report that it experimented with making a 2D game using solely generative AI tools. The result of what is described as an R&D initiative was that those tools are “unable to replace talent”.
The report quotes part of the findings as “whilst GenAI may simplify or accelerate certain processes, the best results and quality needed can only be achieved by experts in their field utilising GenAI as a new, powerful tool in their creative process”. But if you’ve been following the scene, you would know that it was a forgone conclusion. For every claim of the tech taking people’s jobs, there has been equal pushback, saying that it will help people with their jobs instead of replacing them outright.
Of course, there are those who have tried to go down that route for real rather than just testing the waters, with predictable results. Infamously, CNET comes immediately to mind, with the story now so well known it should not need further elaboration. What does, though, is the fact that the flop was so severe that Wikipedia is now lowering the reliability rating of the site for articles after that episode.
Yes, it’s Wikipedia, and as much as it is a student’s greatest saviour, it’s not much seen in a favourable light after. So feel free to make of that what you will.
You probably remember that time when we gave it a go ourselves – on this very subject matter at that – with impressive-looking results that don’t hold up to scrutiny. Sure, the wall of text is impressive if word count is what you’re after, but that’s not difficult to achieve if you don’t mind repeating the same points over and over. Not to mention the questionable accuracy in terms of facts. Which, again, can give the illusion of a profound understanding of a subject, until an actual subject matter expert steps in and takes a look.
Sure, if you’re lucky, generative AI can provide you with a usable skeleton, something that you can work with, with minimal cleaning up. Ideally, that’s the case, anyway, and it will be a while before we get there, let alone the tool being good enough to replace people. But on the flip side, and as it is now, depending on the amount of cleaning up required, it may be more effort than just creating something from scratch, especially where writing an article is concerned.
And it’s not just generated text either. As bountiful as the tools for image generation we have right now, you’ll be lucky to generate the image that you want from a limited number of prompts. More often than not, a desired outcome (impressive enough to win awards) is preceded by countless prompt engineering, or selecting the closest one among the many rejects and applying manual Photoshopping skills. For the former route especially, it may be faster to simply design a desired graphic from scratch.
The same can be said of AI-generated video and audio, but one can imagine that the corrective process here is exponentially longer. That being said, AI text-to-speech tools have been around for quite awhile, and used by those with vocal chord damage, or as another layer of anonymity for content creators. But with such tools being brought more into the limelight, there is an equivalent increase in scrutiny in their misuse. The video game titled The Finals come to mind, with its own flavour of controversy. Though whether or not the AI voices used in the game are good enough to replace actual voice actors, your mileage may vary.
So generative AI tools’ ability to replace humans is questionable at best, but are there examples of it helping people with their jobs? Kind of, and video games have been benefitting from it for quite awhile. Specifically, tools like AI upscaling and frame generation. Of course, it’s not tech that’s exclusive to video games, and the latter can be used for videos in general, and the former can be used for both video and audio.
Ironically, text is probably where generative AI tools are most iffy, even when it comes to simply assisting a human writer. Just about every company announcing their own generative AI tool have tried to sell it by saying that these tools can help you quickly summarise multi-page documents in seconds. Which is nice when it works, but in reality, chances are you’d still put this in the “your mileage may vary” corner. And we’ve already talked about what it’s like letting it take over a person’s writing job.
At current, as much as some may want it to be, AI tools still have a long way to go before they can replace humans in non-menial tasks. But will we be waiting long before that actually happens? That probably depends on your level of optimism. I’m not a very optimistic person, so I’d say we’re still decades away. But Nvidia CEO Jensen Huang says that some form of artificial general intelligence (AGI) will arrive within five years.
This sounds highly optimistic considering we haven’t gotten simpler, much narrower forms of AI up to par. But Huang is someone who definitely knows more about AI than I do, so he may be on to something.
And on that bombshell…
Follow us on Instagram, Facebook, Twitter or Telegram for more updates and breaking news.
