3 ms·

If the generated content is vetted by an adversarial network, we theoretically have a recursively improving AI. Perhaps by introducing more randomness we could

If the generated content is vetted by an adversarial network, we theoretically have a recursively improving AI. Perhaps by introducing more randomness we could even reach some form of evolution that can introduce new concepts, since limited scopes is still what gives AIs away in the end. That is the optimistic perspective at least. On the net, search and content quality is already pretty low for certain key words. If a word is part of the news cycle, expect hundreds of badly researched news paper articles, some of which might be generated as well. Or if they weren't, you wouldn't notice a difference if they become that. But I don't believe the companies made a mistake. They could even protect their position with the data they already acquired and classified. Maybe a quality label would say genuine human®. If all that fails large companies would also be able to employ thousands of low wage workers to classify new content. The increasing memory problem persists, I think that is a race where the model that can extract data as efficiently as possible will win. But without the data sets, there is no way to verify performance.

deeviant 3y ago
This is exactly what I thought when I read the article. Specifically, that it's exactly opposite-of-correct; this is exactly the thing that will push LLMs to the next level. I think the simple idea using chatGPT to filter it's input set with result in drastically higher quality dataset. Add in the ability for chatgpt to use plugins and such like wolfram alpha to fact check it's answers then chatgpt can actually start generating input data from know good sources in areas it knows it's having quality issues. I mean, it literally looks like the beginning of self-learning AI.
sesteel 3y ago
Yep! The singularity concept people talk about relies on a feedback loop like you described. So, she isn't wrong, per se, but just telling a fragment of the story that exist today.

deeviant 3y ago

This is exactly what I thought when I read the article. Specifically, that it's exactly opposite-of-correct; this is exactly the thing that will push LLMs to the next level. I think the simple idea using chatGPT to filter it's input set with result in drastically higher quality dataset. Add in the ability for chatgpt to use plugins and such like wolfram alpha to fact check it's answers then chatgpt can actually start generating input data from know good sources in areas it knows it's having quality issues. I mean, it literally looks like the beginning of self-learning AI.

sesteel 3y ago

Yep! The singularity concept people talk about relies on a feedback loop like you described. So, she isn't wrong, per se, but just telling a fragment of the story that exist today.