People are talking a lot about AI these days. What AI is good for, where it is not wanted, and whether more laws are needed to control it.
It probably depends on how you feel about whether you’ll be happy or horrified to learn how much of what’s on the Internet is made by computers rather than humans.
Deputy We recently reported on a study showing that a “shocking” amount of the web is being generated by low-quality AI.
According to the researchers, more than half of all text on the Internet (57.1% to be exact) is translated into two or more other languages.
The quality of the translation is poor, suggesting that AI using large-scale language models (LLM) was used to create this material. and Do the translation.
And the more vague the language, the worse the translation. This makes sense, given that AI needs data to train itself.
They believe that AI is being used to generate large amounts of English-language content in order to post clickbait for advertising revenue. Then translate it (poorly) into other languages until parts of the web are completely filled with scrambled copies of the copies that still make money.
“Machine-generated multidirectional translations not only account for the total amount of translated content in low-resource languages on the web, but also represent a large portion of all web content in those languages.
Both Amazon and Google will tell you this is nothing new. That’s because both companies are already struggling to cope with AI-generated content littering their sites and search results.
Although this problem exists for English speakers, it is more pressing and prevalent for speakers of other languages.
This could lead to an existential crisis for AI, as AI requires high-quality data to learn and grow. This data is typically collected from the web, but from now on we will only have low-quality information already generated by AI.
I’m not the wiser about what will happen if this turns into a crappy feedback loop.
But I have to think that’s not good.
If you liked this story, check out what happened when a guy gave ChatGPT $100 to make as much money as possible. And it turned out exactly as you expected.
