HN Explorer

Tell HN: I am self-censoring on social media, to avoid AI

7 points by xk_id ↗ HN
There was always the implicit knowledge that all content posted on social media platforms is privately stored by corporations. However, there is now the additional incentive to actively scan and process this content to train AI language models.

I’ve noticed lately in myself that this new perception makes me more self-conscious than before, to the point where I am self-censoring and avoiding to post in my natural, individual “voice”. My unique linguistic character is something I don’t really wish to be modelled and subsequently plagiarised by arbitrary corporations.

This is in addition to prior concerns of fingerprinting the writing style, which presumably is going to become more and more precise, and enables correlating “alt” accounts, perhaps even across platforms.

Technically-inclined people have always recognised those risks. But the success of chatGPT in capturing mainstream’s imagination might make more people aware of this reality. Is this making anyone else reluctant to post original content on social media? Will the quality of online social interactions suffer, because of this suspicion that you risk your “personality” and ideas being analysed, modelled and stolen by the platform you’re using?

7 comments

[ 3.5 ms ] story [ 26.4 ms ] thread
andsoitis
> My unique linguistic character is something I don’t really wish to be modelled and subsequently plagiarised by arbitrary corporations.

How unique do you think is your linguistic character, considering there are 7.888 billion people on the planet?

One perspective is that individuals' traits are but punctuated emergence in the sea of probability.

smoldesu
It just seems like a reductive argument to me. Why bother ever expressing yourself if you're afraid of someone else copying you?

With or without AI, I don't really follow the line of logic.

xk_id
In normal interactions, the expectation is that you are being exposed only to your conversation partner, which you’ve consented to. In online interactions, you’re increasingly being exposed to arbitrary, hidden parties, with the computing power and private incentives to process your conversations. This is a big difference.
smoldesu
Yep. Baring this in mind, not much has changed about the way I behave online.
anigbrowl
HN's corpus was included in OpenAI's training dataset, so it might be tool late for that. You could do what I do and flatter yourself that its impressive abilities are thanks in no small part to the brilliance of your prior contributions, and until they release the weights nobody can prove otherwise.
[dead] spvx
[dead]