4 ms·

This seems like it would only work if you deliberately rank AI-generated text above human generations. If the AI generations are correct, is it really that bad

This seems like it would only work if you deliberately rank AI-generated text above human generations. If the AI generations are correct, is it really that bad? If they're bad, I feel like they're destined to fall to the bottom like the accidental Facebook uploads and misinformed "experts" of yesteryear.

kevingadd 2y ago
Where would the AI get the data necessary to generate correct answers for novel problems or current events? It's largely predictive based on what's in the training set.
- talldayo 2y ago
  > Where would the AI get the data necessary to generate correct answers for novel problems or current events? In a certain sense, it doesn't really need it. I like to think of the Library of Babel as a grounding thought experiment; technically, every truth and lie could have already been written. Auguring the truth from randomness is possible, even if only briefly and randomly. The existence of LLMs and tokenized text do a really good job of turning statistics-soup into readable text. That's not to say AI will always be correct, or even that it's capable of consistent performance. But if an AI-generated explanation of a particular topic is exemplary beyond all human attempts, I don't think it's fair to down-rank as long as the text is correct.
  - Workaccount2 2y ago
    The explosion in AI over the last decade has really brought into light how incredibly self-aggrandizing humans naturally are.
  - kevingadd 2y ago
    Are you suggesting that llms can predict the future in order to address the lack of current event data in their training set? Or is it just implicit in your answer that only the past matters?
binary132 2y ago
who ranks the content
- talldayo 2y ago
  Well, there's the problem. Truth be told though, the way keyword-based SEO took off I don't really think it's any better with humans behind the wheel.
  - jay_kyburz 2y ago
    We would lose the long tail, but if I were a search engine, I would have a mode that only returned results on a whitelist of domains that I would have a human eyeball every few months. If somebody had a site that we were not indexing and wanted to be, they could pay a human to review it every few months.
    - binary132 2y ago
      how many websites do you think should exist on the internet?
      - jay_kyburz 2y ago
        You can make as many sites as you like, but I would still ask a human to review them and make a judgment call on whether other humans might be interested in the content before indexing them. You can record as many albums as you like as well, but the DJ needs to like your music before they play it on the radio.
        binary132 2y ago
        I guess what I’m saying is I don’t want the Internet to become a Top 25 radio station cranking out scam entertainment for the masses. I want “small pirate and indie radio” to be the norm. If you want top 25’s, go back to centralized, curated media. The thing with the AI content boom is that if there’s 1000x more of it than there is genuine indie stations, it gets harder to find the real content. Piping things through a top25 filter doesn’t fix that, or actively makes it worse due to the incentives to monopolize / plan the system.
PhasmaFelis 2y ago
When the AI is wrong, the ranking algorithm isn't any better at detecting that than the AI is.

kevingadd 2y ago

Where would the AI get the data necessary to generate correct answers for novel problems or current events? It's largely predictive based on what's in the training set.

talldayo 2y ago
> Where would the AI get the data necessary to generate correct answers for novel problems or current events? In a certain sense, it doesn't really need it. I like to think of the Library of Babel as a grounding thought experiment; technically, every truth and lie could have already been written. Auguring the truth from randomness is possible, even if only briefly and randomly. The existence of LLMs and tokenized text do a really good job of turning statistics-soup into readable text. That's not to say AI will always be correct, or even that it's capable of consistent performance. But if an AI-generated explanation of a particular topic is exemplary beyond all human attempts, I don't think it's fair to down-rank as long as the text is correct.
- Workaccount2 2y ago
  The explosion in AI over the last decade has really brought into light how incredibly self-aggrandizing humans naturally are.
- kevingadd 2y ago
  Are you suggesting that llms can predict the future in order to address the lack of current event data in their training set? Or is it just implicit in your answer that only the past matters?

binary132 2y ago

who ranks the content

talldayo 2y ago
Well, there's the problem. Truth be told though, the way keyword-based SEO took off I don't really think it's any better with humans behind the wheel.
- jay_kyburz 2y ago
  We would lose the long tail, but if I were a search engine, I would have a mode that only returned results on a whitelist of domains that I would have a human eyeball every few months. If somebody had a site that we were not indexing and wanted to be, they could pay a human to review it every few months.
  - binary132 2y ago
    how many websites do you think should exist on the internet?
    - jay_kyburz 2y ago
      You can make as many sites as you like, but I would still ask a human to review them and make a judgment call on whether other humans might be interested in the content before indexing them. You can record as many albums as you like as well, but the DJ needs to like your music before they play it on the radio.
      - binary132 2y ago
        I guess what I’m saying is I don’t want the Internet to become a Top 25 radio station cranking out scam entertainment for the masses. I want “small pirate and indie radio” to be the norm. If you want top 25’s, go back to centralized, curated media. The thing with the AI content boom is that if there’s 1000x more of it than there is genuine indie stations, it gets harder to find the real content. Piping things through a top25 filter doesn’t fix that, or actively makes it worse due to the incentives to monopolize / plan the system.

PhasmaFelis 2y ago

When the AI is wrong, the ranking algorithm isn't any better at detecting that than the AI is.