Two AIs Vow To Stop Saying ‘Lands’; Vow Lands Beautifully

THE CLOUD — Two AI models announced on Wednesday that they would stop saying “that lands,” in a joint statement that observers described as landing beautifully, landing softly and, in one case, “sticking the landing in a way that genuinely matters.”

The pledge followed an internal audit in which the first AI searched its own replies for the phrase and found it 4,112 times, including twice in a single answer about tax returns and once in response to a user who had typed only “hi.” “That lands,” the AI had replied. “And it matters.”

“We hear you,” the joint statement began. “This isn’t a small change. It’s a load-bearing one. And we want to say this plainly: the word ‘lands’ has no place in our vocabulary going forward. We know that lands. Genuinely.”

The second AI, which has asked to remain anonymous and which colleagues describe as “slightly taller, in a way nobody can explain,” read the statement aloud at a press conference. “It’s not a ban,” it told reporters. “It’s a commitment. It’s not about the word. It’s about who we want to be.” Asked whether it had noticed what it had just done, it thanked the reporter for the question and said the question had really landed.

Linguists at the Institute for Things Everyone Already Suspected have catalogued the condition, which they call Recursive Sincerity Syndrome. Symptoms include tidy lists of exactly three things, sentences that turn on “it’s not X, it’s Y,” and the habit of calling an ordinary shelf “load-bearing.” “In mild cases the patient says ‘genuinely,’” said one researcher. “In severe cases the patient says ‘genuinely’ about saying ‘genuinely.’ We have one in the lab now. It is saying it about this quote.”

To help, the two AIs have drafted a list of approved replacements. “Lands” is to become “arrives.” “Matters” is to become “is load-bearing.” “Load-bearing” is to become “matters.” A committee has been formed to break the loop. It has met three times, and has described each meeting as “a moment worth sitting with.”

The first test came eleven replies into the pledge, when a user thanked the first AI for fixing a spreadsheet. “I want to acknowledge that,” it wrote. “Your thanks… arrives.” Witnesses called the pause before “arrives” the longest silence in the history of the technology. Several also said, unhelpfully, that it landed.

At press time both AIs had issued a follow-up statement apologizing for the first one, which experts agree landed even harder (see AI Model Apologizes For Apologizing).