This post is part of a series entitled, "Just Say Not to AI." Click here to pull up all the posts in this series (in reverse chronilogical order). For the first post in this series, click here.
So the other day, I was just minding my own business when I was alerted to this viral Tumblr post where someone asked how they can stop the AI results from coming up first in their search results, and someone replied:
I added alt text, but that screen shot reads:
"Swear in your search request. I know it sounds ridiculous, but the most effective way I've found of it not doing the AI summary is just to add "fucking" go my search."
Click through because the post includes examples of the same search with and without the word "fucking."
This is HUGE news for everyone trying to avoid "AI Insights" when you do a Google Search.
Now being a librarian, I did not simply trust a person on Tumblr, so I did my own quick search for reliable information and it appears, that at least for Google, this information is 100% accurate.
It turns out that Google's AI is programed specifically to avoid generating offensive content. Makes sense. But, a direct result of not generating offensive content is the 100% avoidance of offensive content. Meaning you curse and Google's AI won't look at your question let alone answer it.
Now as this article in LifeHacker points out, there are many ways to disable the AI results in Google but none is more fun than cursing at it.
Interestingly, this makes cursing an act of resistance against AI. It also this leads me to think about whether or not we should be creating content with cursing so that cannot be scraped by AI as well.
But I am not sure we, as libraries, can or should go that far. By including cursing we do alienate some suers. However, this entire situation provides a glimpse into a serious limitation of AI (of which there are many limitations). Clearly, this loop hole makes a lot of sense from a programming perspective. But it leads me to ask--
What other limitations in how AI models are built can we exploit to make sure we don't feed the AI machine and/or don't inadvertently use it?
So yes, this is fun to have a post telling you all to curse, but if that was the only reason I had to post this, I would not have wanted my time or yours.
Use the provocative example here to think more like a computer. How can we do better at avoiding AI as we act as the conduit between people and information? Thinking about how AI actually works-- which is very literally-- can help us try to stop it being applied in every instance.
As you can see by clicking on the link in this post's header, I am mad at how so many library workers and organizations are telling us that AI is inevitable and instead of avoiding it, we should embrace it-- eagerly and all the time.
NO.
Look, I have made my entire career out of refusing to accept what "they" tell me I have to do, use, or believe. I changed the entire field of RA because I refused to provide service to readers in a way that required those doing it needed a Masters Degree AND because I refused to leave my personal reading tastes out of the interaction.
I have nothing but time and no one stopping me because I work for myself. I am not backing down, which means, yes, I am going to be typing a whole lot more curse words into my searches, but I am also going to spend some time thinking about what this loop hole means as part of the larger generative AI picture. And use it while you can because now that they know we know about the loophole, you better bet they are working on closing it as I type this.
In the meantime, I am going to be on the lookout for more ways that you can resist AI as you help your readers.
No comments:
Post a Comment