ChatGpt, Gemini, and other generator AI products have applications, but some companies are outboard. Beyond issues like hallucinations and removing the entire code database due to AI being “panic”, there are also concerns about how AI is being used without user knowledge or permission. YouTube has given us the perfect example of how it happens.
In one of the platform's latest experiments, YouTube began small editing of several videos without first warning the creator. The change was not made by the generator AI, but it relied on machine learning. In most cases, the reported changes appear to add definitions to things like wrinkles.
YouTube has implemented useful AI tools in the past, but these latest changes are part of the bigger issue, such as helping creators come up with video ideas. Made without the consent of the user.
Why is consent important?
We live in a world where AI is increasingly inevitable due to lack of regulations. Officials like President Trump are unlikely to change anytime soon as they continue to push for AI action plans to help businesses invest in and expand it as quickly as possible. Therefore, it is up to these companies to prioritize seeking consent from users when implementing AI.
According to a BBC report, some YouTubers are more interested than others. For example, YouTuber Rhett Shull created an entire video that focused attention on YouTube's AI experiments. YouTube addressed the experiment a few days ago, and YouTube creator Liaison Rene Ritchie points out in X that this is not a result of the generation AI. Instead, machine learning is used to “amble, remove, and improve clarity of video during processing (as is what modern smartphones do when recording videos).
YouTube has a great deal of control over all the content users upload. it doesn't matter. The problem is the fact that YouTube is doing this without the user's consent. This is because this means that these videos are treated as training materials for the machine learning process. And it has always been an AI development issue.
Machine learning is still AI
Generic AI is certainly the industry talk now, but machine learning is still AI. Behind the scenes there is still an algorithm that does all the heavy lifting, and it works with the material being trained. YouTube can identify machine learning with the same thing as smartphone cameras, but the difference here is that you know that mobile phones do it. YouTube didn't reveal the existence of this experiment until someone started complaining about it.
It's not the right way to handle AI, especially since it's far from perfect. Machine learning may not suffer from the same pitfalls as generator AI, but just because YouTube doesn't have to worry about us taking crime alerts that created foing bug AI like other apps, this is not about the company making this an invasive move and continuing AI as much as possible.
YouTube doesn't share when the experiment ends or if there will be a wider development in the end. That being said, if you're watching YouTube shorts and notice that the videos look a bit weird and weirdly rising, it's probably because YouTube started editing those videos and even pissing off some people, they've tried to make them better in some way.
