I posted this on Mastodon, showing Perplexity, which is a search engine that says it will browse the web for you and give you the answer, just straight up plagiarizing the first result for the question I asked. And I’m not talking plagiarism like the adamantly anti-LLM crowd thinks everything in LLM does is plagiarism. No, I mean literally copy/pasting and changing a few words to make it look like it’s your own. The cherry on top for me was the fact that Perplexity cites its answers, showing you where it got that part of its answer, and in this case it’s the same website for every single step of the process. Awesome job guys great product. 🤥

So you might reasonably ask yourself, "Matt, if you’re upset about this, then why are you supportive of LLM’s at all?" Well, just like with all technology, there are good ways to use the tech and bad ways to use it.

Look at some other examples for a comparison. You could read this very article and use the tools on your computer to copy the text, paste it into your own web CMS, and publish it as your own creation. We can all agree that that’s wrong and you shouldn’t do that, but we wouldn’t argue that the copy and paste functions of computers should not exist and not be used. Similarly, I could download a paid wallpaper pack which delivers me a JPEG file, and we can all pretty much agree that if I started sharing those JPEG’s for free with anyone who wanted them that would be wrong as well. Are we suggesting that JPEG’s should not be allowed to be copied? I don’t think so.

And in much the same way, I think LLM’s have good uses and bad uses in my eyes. The good uses are the things where an LLM helps you do something that you are already working on. So for example, helping you proofread your writing is useful. Helping you check the tone of something you’ve written is useful. The ability to extract useful information from a PDF that you don’t understand is useful. Being able to reword complex things that you don’t grok are useful!

What I don’t like are the things that Perplexity, Arc Search, and increasingly Google are doing where they slurp up content (copy) and then slightly reword it as their own (paste).

I understand the argument that because LLM’s can be used for bad, we can’t enjoy the good and I would just say that’s the same with all technology; we just have rules and norms around how we handle all of those and I think we’re going to have to adopt similar things for LLM’s.