Innovation starts with consumers, not academia

I don't believe it -- and I expect better from Lemire.

My understanding is that a lot of consumers are really pissed that industry is forcing AI products down their throats. I mean, I like Microsoft Copilot but a lot of people don't -- and I haven't heard a single good thing about the hallucinated trash results that Google adds to search without our consent.

When it comes to LLMs there were people doing abstract work on neural networks for sequences in the 1980s in academia which really didn't go anywhere, lots of work on language modelling with markov chains and conditional random fields, Hinton's work on deep learning in the mid-00's and then in the mid-10's I worked on

-- a deep learning search engine for patents (big success!)

-- an attempt to do language modelling of clinical notes based on LSTMs

-- foundation models based on CNNs for text classification (startup got bought by a leading fashion brand)

all industrial activity predating consumers having any idea any of this was possible.