·
I was thinking recently about how we stopped using the word NFT (while continuing to use the same tools), and today of course, we have another overused word that is starting to lose its meaning.
On my recent trip to China, I was a bit shocked to see this familiar word on every second billboard. These were the only symbols I could understand, not knowing Chinese, and they were looking at me from every corner.
As with all bubbles, this word will probably soon become tired, acquire a bad reputation, and we'll start calling it something else. Especially given that the word itself is fairly meaningless: from its first use at the 1956 Dartmouth Conference, to the multilayer perceptron, which has little in common with modern transformer-based architectures, to chatbots, to data science and simple correlation.
Do you see any change in vocabulary when talking about AI, and have you perhaps started replacing the word AI with other words when you speak?
4 comments
AI? IA! IA! Cthulhu fhtagn! Ph'nglui mglw’nafh Cthulhu R'lyeh wgah'nagl fhtagn!
;-)
but seriously, I actually have always tried to avoid the term “AI” because it’s always been very ill-defined. even before LLMs. I used to study Machine Learning (I stopped shortly before neural networks got cool again, though), and we never really called it “AI”, because what really is, AI? so I always try to be a bit more exact than that, if I know something to be “computer vision”, “LLM”, “diffusion style image generation”, “recommendation algorithm”, etc
if I just said “AI” it’d be synonymous with “computer doing thing I don’t understand”, and if that is the case, I’d rather just say “I don’t know what it’s doing” :-)
one exception I think, in video games if you’re playing against a computer opponent it makes sense to call that an “AI” (although ... it’s still quite interesting to learn about A*, or alpha beta minmax pruning)
thanks for sharing that! I agree, I have actually not used the term much before either. it was deep learning mostly for me.
and AI as an umbrella term for everything from very naive methods (like regression) to everything else.