Lea Verou on Nostr: [2/2] In practice, it’s hardly ever right, so it just adds friction. It assumes ...
[2/2]
In practice, it’s hardly ever right, so it just adds friction. It assumes most emoji usage is pictorial duplication of meaning, rather than than *adding* meaning, which I’d wager is more common but harder to predict.
I suspect it varies by demographic too — but a good implementation would *monitor* manual usage and use it to train the model to do better in the future.
In practice, it’s hardly ever right, so it just adds friction. It assumes most emoji usage is pictorial duplication of meaning, rather than than *adding* meaning, which I’d wager is more common but harder to predict.
I suspect it varies by demographic too — but a good implementation would *monitor* manual usage and use it to train the model to do better in the future.