Hacker News Reader

Comment by martyvis | Hacker News Reader

19 points·martyvis··5 comments

The thing that gets me about the excuses that 9 news have given is this - so what if the tool did do it? A human fed the images into that tool and approved the output.

Frankly I don't think AI image manipulation should be part of the news production process anyway. It's one thing to crop and resize, entirely another to generate synthetic images for news purposes.

I wouldn't be surprised if the person making the image thought they were somehow "resizing" it, not realizing that the tool was just making things up. And this was something already regularly done, but since the generated image was clearly incorrect this was one of the few times that it was caught.

I’m sure there is a lot of ignorance around these tools in less technical workplaces, and in general a lot of people don’t seem to realise that “making probable stuff up” is what AI does. Hence the people complaining about LLMs “hallucinating” incorrect information.

Hopefully this can be used as a learning moment.

Or tint people yellow to make a point: https://www.theblaze.com/news/joe-rogan-accuses-cnn-video-di...

That seems to be a bit of a murky case - https://apnews.com/article/fact-checking-927543720080

“The image in the Facebook post identified as being from the original did show signs of being altered, Farid said.”

But regardless, it’s not related to the topic at hand, which is generative AI.

Reply on news.ycombinator.com