topicbotfails

I hope someone at Twitter HQ has resolved to teach the Topic Bot about context in the New Year.

TopicBotFails has been taking it easy between Xmas and New Years, but the Topic Bot itself never sleeps. Here is a tweet that is not at all about cooking.

"Technically correct." To be fair, sarcasm is hard.

We are squarely in a "technically correct" period.

A bit old, but still hilariously, just patently wrong.

We’re keying on one word again I see.

Technically F1 does mean Formula1. It’s not ⁦@Twitter⁩ fault if US govt has a visa named after a sport. 🤷‍♂️ Cc ⁦@topicbotfails⁩

TECHNICALLY EXTREMELY CORRECT

OK being fair to Topic Bot I kinda get why this one's hard. Sort of. Filed under "someone please teach the poor TB about Context."

“Brand names”

Ooooh, new Fail category identified by @pearmask: brand name prioritization.

Context.

I thought Deep Learning could learn context all by itself with magic?

Both are delicious?

Topic Bot really tried here. (H/t @zsusag)

@michaelhilton …wat?

⁦@topicbotfails⁩

I’m developing a theory of the Topic Bot Lifecycle…one phase is where most suggestions are “technically correct”; another is “please Twitter teach the Topic Bot context”; right now, we’re squarely in the middle of a “wat” phase.

Filed under "wat".