We had predictive text. We had neural networks. We had both for more than few decades. Someone just decides “what if predictive text was built on neural network?” and it worked much better than old solution. In fact old solution worked pretty much like one-layer neural network.
TechBros literally took something we already had, made it more expensive and sold it to investors.
ehh not exactly. the current wave within ai was started in 2017 with the google paper “attention is all you need”, which introduced the transformer architecture (the T in GPT). deep neural networks were still used for NLP before this, it was just LSTMs instead of transformers, which had various limitations
current wave of AI isn’t even new technology
We had predictive text. We had neural networks. We had both for more than few decades. Someone just decides “what if predictive text was built on neural network?” and it worked much better than old solution. In fact old solution worked pretty much like one-layer neural network.
TechBros literally took something we already had, made it more expensive and sold it to investors.
ehh not exactly. the current wave within ai was started in 2017 with the google paper “attention is all you need”, which introduced the transformer architecture (the T in GPT). deep neural networks were still used for NLP before this, it was just LSTMs instead of transformers, which had various limitations