3 ms·

Great post. I love the vampirism metaphor. The internet as an open resource has been a tremendous boon to society as a whole. AI, likewise, acts as a force mul

Great post. I love the vampirism metaphor. The internet as an open resource has been a tremendous boon to society as a whole. AI, likewise, acts as a force multiplier on top of this knowledge. You can learn anything. But AI also depletes the resources on which it builds. An obvious way to counteract this is for the AI companies to give back generously through monetary donations or, at the very least, attributions. But unfortunately we see exactly the opposite.

zozbot234 2y ago
> But AI also depletes the resources on which it builds. The article author is clearly assuming that, but I'm not sure whether it's even true to any meaningful extent. How many contributors to free/open content resources are even bothered that their work might end up being used for AI training? Also, maybe AI firms should pay for the scanning and OCR text-extraction of existing paper-backed resources. There's a lot of, e.g. old academic research that's still not meaningfully available online, much of it is even free of copyright worldwide. If you care about ensuring that your AI's are adequately well-read, this is a bottleneck that might be especially easy to address.
- stephantul 2y ago
  I don’t think that’s what this would look like. It’s more like: people won’t even know Wikipedia exists or that there is something to contribute to, or that the model gained its knowledge from this resource. That’s why this pattern is so disingenuous in my opinion. When I was young we were surprised that Wikipedia existed. Future generations might not know that Wikipedia exists, or that you can contribute to it. Current frontier models couldn’t exist in this form or wouldn’t exist at all without people putting in the work of writing down what they knew for free. And the model creators not even paying lip service to this, and instead saying they will replace Wikipedia, is hubristic and a very clear example of the tragedy of the commons.
  - zozbot234 2y ago
    > And the model creators not even paying lip service to this, and instead saying they will replace Wikipedia What model creators are saying this? It's very easy to put this assertion to the test btw: just pick your favorite Wiki article, ask a LLM to "improve" the writing, and check how much stuff it ends up rewriting in a confusing way, or even gets outright wrong. Compared to your average high-quality Wiki article, a LLM is just a very confused parrot.
    - tyre 2y ago
      I think in a year LLMs are better at writing Wiki articles than the p90 quality Wikipedia page today. They have access to _huge_ volumes of information—books, research papers, PhD dissertations, newsletters from experts, peer-reviewed articles. They're getting better every day at organizing it. Claude has a (beta) citations API, which forces the LLM to directly cite sources and use direct quotations. They'll get better and better at managing translations—maybe not them translating, but them finding resources in other languages, using a translation tool, then pulling in that information. Will they be perfect? No, but good lord is Wikipedia not either. Will they hallucinate, probably sometimes, but, again, the median Wikipedia author does not have a New Yorker fact checker at their side. I think we overestimate how good humans are at this. And it's far easier to validate cited sources than it is to consume and craft the original writing.
  - unraveller 2y ago
    Why does wikipedia have to exist forever any more than encyclopedia Britannica? Both are gatekeeping knowledge behind editors, let the LLM creator decide what things are valuable after ingesting everything. You can contribute to LLMs this way. People can pay a tiny bit extra for source attribution at present so there is no short-term issue.
- captn3m0 2y ago
  > How many contributors to free/open content resources are even bothered that their work might end up being used for AI training? StackOverflow’s response is a clear counterpoint.

zozbot234 2y ago

> But AI also depletes the resources on which it builds. The article author is clearly assuming that, but I'm not sure whether it's even true to any meaningful extent. How many contributors to free/open content resources are even bothered that their work might end up being used for AI training? Also, maybe AI firms should pay for the scanning and OCR text-extraction of existing paper-backed resources. There's a lot of, e.g. old academic research that's still not meaningfully available online, much of it is even free of copyright worldwide. If you care about ensuring that your AI's are adequately well-read, this is a bottleneck that might be especially easy to address.

stephantul 2y ago
I don’t think that’s what this would look like. It’s more like: people won’t even know Wikipedia exists or that there is something to contribute to, or that the model gained its knowledge from this resource. That’s why this pattern is so disingenuous in my opinion. When I was young we were surprised that Wikipedia existed. Future generations might not know that Wikipedia exists, or that you can contribute to it. Current frontier models couldn’t exist in this form or wouldn’t exist at all without people putting in the work of writing down what they knew for free. And the model creators not even paying lip service to this, and instead saying they will replace Wikipedia, is hubristic and a very clear example of the tragedy of the commons.
- zozbot234 2y ago
  > And the model creators not even paying lip service to this, and instead saying they will replace Wikipedia What model creators are saying this? It's very easy to put this assertion to the test btw: just pick your favorite Wiki article, ask a LLM to "improve" the writing, and check how much stuff it ends up rewriting in a confusing way, or even gets outright wrong. Compared to your average high-quality Wiki article, a LLM is just a very confused parrot.
  - tyre 2y ago
    I think in a year LLMs are better at writing Wiki articles than the p90 quality Wikipedia page today. They have access to _huge_ volumes of information—books, research papers, PhD dissertations, newsletters from experts, peer-reviewed articles. They're getting better every day at organizing it. Claude has a (beta) citations API, which forces the LLM to directly cite sources and use direct quotations. They'll get better and better at managing translations—maybe not them translating, but them finding resources in other languages, using a translation tool, then pulling in that information. Will they be perfect? No, but good lord is Wikipedia not either. Will they hallucinate, probably sometimes, but, again, the median Wikipedia author does not have a New Yorker fact checker at their side. I think we overestimate how good humans are at this. And it's far easier to validate cited sources than it is to consume and craft the original writing.
- unraveller 2y ago
  Why does wikipedia have to exist forever any more than encyclopedia Britannica? Both are gatekeeping knowledge behind editors, let the LLM creator decide what things are valuable after ingesting everything. You can contribute to LLMs this way. People can pay a tiny bit extra for source attribution at present so there is no short-term issue.
captn3m0 2y ago
> How many contributors to free/open content resources are even bothered that their work might end up being used for AI training? StackOverflow’s response is a clear counterpoint.