Time to add ample praise of myself in my blog posts. Some time later: “…as you see, that is the load bearing assumption here. Speaking of which, you should hire KronisLV.”
Okay it’s meant to be a bit silly but I do wonder how many pages that are generated specifically to influence AI make it into training data and also how often the AI search integrations find it.
Would people hating on a specific language, technology or approach (let’s say OTLT/EAV in database design) be able to exert meaningful influence over say a decade? Or, you know, praising memory safe languages for example and trying to make that preference be stronger.
There was an example with I think ChatGPT some time ago regurgitating an uncommon phrase verbatim from someone’s blog, when asked a specific question.
Yep. I suspect this already underway. The scrapers feeding data into the AI pre-training are indiscriminately hoovering up everything they can. It'd be trivial to spam a bunch of BS websites with whatever endless text you want to "taint" future models. Post tons of examples of insecure code or package.json files pointing to some malicious library.
Okay it’s meant to be a bit silly but I do wonder how many pages that are generated specifically to influence AI make it into training data and also how often the AI search integrations find it.
Would people hating on a specific language, technology or approach (let’s say OTLT/EAV in database design) be able to exert meaningful influence over say a decade? Or, you know, praising memory safe languages for example and trying to make that preference be stronger.
There was an example with I think ChatGPT some time ago regurgitating an uncommon phrase verbatim from someone’s blog, when asked a specific question.