Another very important reason why I prefer Perplexity is due to attribution. It actually does cite the sources that it bases it's output on (unless it's something generic or calculated), so if I suspect something is off or want to look deeper into some particular aspect I can easily click through. And I've done enough click-throughs to be confident that Perplexity faithfully represents sourced content, and accurately gets exactly the bits I'm interested in, maybe 98% or more of the time.
It is your prerogative to tune your servers as you see fit, but as LLM adoption increases you'll merely find that your site has fewer and fewer visits overall, so your content will only be utilized by you and a vanishingly small group of other persons. Perhaps you're OK with that, and that's also fine for the rest of us.
It's strange you mention theft, and then say it isn't about money. For me, and many others, it's about practicality and efficiency. We went from having to visit physical libraries to using search engines, and now we're entering the era of increasingly intelligent content fetch+preprocess tools.
> as LLM adoption increases you'll merely find that your site has fewer and fewer visits overall, so your content will only be utilized by you and a vanishingly small group of other persons.
So far, AI has had the opposite effect on my site. I've now been featured on both Hackaday and Adafruit's blog. Both features were clearly AI-generated. Both posts coincided with an influx of emails from folks interested in my work.
Perplexity is good at citing things when it decides to cite things and when you tell it to cite things. It can and does spit out plain expository text with no indication of the information's origin. I do appreciate that you have better-than-usual habits about validating sources.
I think you may have misinterpreted my remark about money. With the direction conversations around AI have been going lately, I was expecting a backhanded accusation that I was farming ad revenue.
"It's not about money" meant that I have nothing to lose financially by losing direct human traffic to my websites. Instead, I stand to lose those aforementioned email conversations.
> So far, AI has had the opposite effect on my site. I've now been featured on both Hackaday and Adafruit's blog. Both features were clearly AI-generated. Both posts coincided with an influx of emails from folks interested in my work.
This may be missing some context, but it seems as though you're saying that you made something with AI and it led to traction. That's great! Seems off the point that blocking LLM service will lead to less exposure over time though.
> Perplexity is good at citing things when it decides to cite things and when you tell it to cite things.
Maybe I'm just lucky, but a quick skim of my Perplexity history yielded only 2 instances of no citations, and they were for general coding queries. I've never had to ask it to cite anything, as that's built into the default prompt.
> lose those aforementioned email conversations.
I think those will remain a possibility as long as LLM users, or services, ensure citations are included in output.
> This may be missing some context, but it seems as though you're saying that you made something with AI and it led to traction. That's great! Seems off the point that blocking LLM service will lead to less exposure over time though.
Hah, I can see how you would have read it that way. Quite the opposite. I don't use AI tools for my writing. Hackaday and Adafruit have both featured my posts, and their posts were pretty clearly AI-generated.
That still sounds like a great deal. Less work for those post authors, and you benefitting from being cited in some way (maybe they used Perplexity or similar, and didn't even visit your site themselves).
@ryukoposting - i am the founder of hackaday, but do not run the site now, and i am also the managing director of adafruit and editor of the adafruit blog. the adafruit does not use generative text, etc. unless clearly indicated https://www.adafruit.com/editorialstandards ... appreciate a correction to your post, hard to combat misinformation with ai, but you can email and i can prove i am human if ya want... pt at adafruit dot com
It is your prerogative to tune your servers as you see fit, but as LLM adoption increases you'll merely find that your site has fewer and fewer visits overall, so your content will only be utilized by you and a vanishingly small group of other persons. Perhaps you're OK with that, and that's also fine for the rest of us.
It's strange you mention theft, and then say it isn't about money. For me, and many others, it's about practicality and efficiency. We went from having to visit physical libraries to using search engines, and now we're entering the era of increasingly intelligent content fetch+preprocess tools.