Why did autocorrect (and swipe predictions) seem to go to shit around 2019? I remember thinking it had gotten really good… and then it got gradually worse. It’s not AI, I think it started going downhill before that, but why? I can’t see a profit incentive and it didn’t make digital communication sufficiently more difficult for conspiracy-sounding reasons.
I feel like it’s just a case of “we must constantly iterate on everything.” So they started adding more and more “predictive” nonsense because where else do you go with autocorrect aside from the branch that is LLMs?
But then the predictive text was basically super duper light LLM. “Autofill the next most likely word or phrase” without realizing the nuance every person has to their phrase patterns.
Why did autocorrect (and swipe predictions) seem to go to shit around 2019? I remember thinking it had gotten really good… and then it got gradually worse. It’s not AI, I think it started going downhill before that, but why? I can’t see a profit incentive and it didn’t make digital communication sufficiently more difficult for conspiracy-sounding reasons.
I feel like it’s just a case of “we must constantly iterate on everything.” So they started adding more and more “predictive” nonsense because where else do you go with autocorrect aside from the branch that is LLMs?
But then the predictive text was basically super duper light LLM. “Autofill the next most likely word or phrase” without realizing the nuance every person has to their phrase patterns.