Discussion
"The media has largely let [tech companies] set the terms of the debate, right down to the terminology used in any discussion of these systems."
From Nanna Inie and me in Tech Policy Press on how to spot and resist anthropomorphizing language in the discourse about so-called "AI".
https://www.techpolicy.press/we-need-to-talk-about-how-we-talk-about-ai/
@emilymbender This is indeed a very valid point. And the other thing I'd love to see prioritised in articles about AI is the extreme impact on energy usage.
For example there was an article recently on The Verge talking about people being unable to identify AI generated music
https://www.theverge.com/entertainment/830393/ai-music-deezer-survey-spotify
In which they talk about 50,000 100% AI songs being uploaded **every** day.
At no point do they point out the ridiculous energy wastage being caused by this (at every point in the process)
I've been sharing this to relatives and presented it as "the AI-error".
(or maybe AI-errand ?)
It is that of the media but also that of managers!
The other day the graphic-card maker talked about physAI that understands... it's really the same error and realizing this is so important to turn back tools to... #instruments (in the sense of the instrumental-approach at least).
@emilymbender an excellent and professional stance! Far far away from the revolutionary tone of #antiAI or #mundaneslop language that a lot of the web has felt the need to use.
FWIW the word agent has been in the AI textbooks since decades. So, indeed, it was just a matter of picking up. Similarly the humanization of computers has been in the study since a bit (e.g. "The media equation").
What I like most is your description of language spread: it makes no one centrally responsible! Just, all!
@emilymbender I think chatbots using the first person singular has been the worst part.
@emilymbender this is where "AI slop" is such a great phrase, and it's important that we keep using it.
@emilymbender
Great article Prof. Bender ;)
We should have called it Artificial Text Generator instead of AI.
@emilymbender I wish they'd proposed more examples of alternative language
@emilymbender Thank you for the article! I am afraid we are already at a level, where sycophancy and self-affirmation mechanisms are built in on purpose at the chat interfaces of LLMs. People get, by design choices, emotionally bound to the product. That is despicable and needs to be regulated.
