If this was a mistake, the proper response might have been "sorry, we applied automation in error, those changes have been rolled back while we fix the process that allowed it to happen". And not "call with us to talk about this further".

At the very least, ‘we will discuss how this could of been handled differently’, not I’m sorry you feel this way.

Mozilla has offered to call the OP, too. I’m curious on the outcome.

They said sorry for how you feel about it which is insincere and unhelpful.

My partner has been picking me up on the specifics of wording.

Is there a slightly different phrasing that would make this better, or is it the sentiment that's crap?

"I'm sorry for how these changes impacted you"? Personally just the sentiment feels insincere to me haha.

The sentiment is more important. But I'm sorry for how you feel suggests to many people the sole problem was their feelings. I'm sorry for how these changes impacted you suggests the changes could have been wrong.

I don't think it's the specific phrasing. They could have said "I'll contact you by email to try and understand your concerns" and it's still dodging the explicit, concrete list of grievances.

However, "let's hop on a call" is just additionally dismissive.

The response was likely also written by AI so there is no point analyzing it. It just ads insult to injury.

Two things stand out, besides what has been already mentioned.

* The infantile corporate-cutesy wording "hop on a call" is not appropriate when talking to somebody who feels that you deeply wronged them. It has the same vibes as cheery "Remember: At Juicero, we are all one big family!" signatures on termination notices, and Corporate Memphis.

* In the first sentence, Kiki says "about the MT workflow that we just recently introduced". Why is this level of detail shoehorned in? Everyone in that conversation already knows what it is about. It's as if Kiki can't resist the temptation to inject an ad/brag about their recently introduced workflow for any drive-by readers. "I'm sorry you were dissatisfied with your Apple(R) iPlunger X(TM), which is now available at major retailers for only $599!"

Don’t use passive voice in an apology. “We’re sorry that we made the change without consulting your team or considering your circumstances.”

The change did not fall out of thin air. It was something they did. If they do not own it explicitly then it’s insincere full stop.

How?

They don't know what exactly has gone wrong. All they can say sorry for is for how the person is feeling. Then they want to get on a call to learn more. Which is the start of helping.

The response is as sincere and helpful as it could be for an initial response from someone who wants to figure out what the problem is.

But he lists the problems? Pretty unambiguously.

The problems are nowhere near actionable. A lot more information is needed.

E.g. literally the first bullet: "It doesn't follow our translation guidelines". OK -- where are those guidelines? Is there a way to get it to follow them, like another commenter says works? Does the person need help following the process for that? Or is there a bug? Etc.

These are the things a call can clarify. It's the necessary first step, so why are people complaining?

> where are those guidelines?

It's entirely possible that such information is well-known to everyone involved in the translation community.

I would consider it outright insulting if someone who ostensibly "wants to help" doesn't know basic information like that - if the people making decisions about SumoBot are NOT aware of basic information like "where to find the local translation guidelines" then they are presumably not qualified to release a tool like SumoBot in the first place.

Yep agree with this. Nothing is more infuriating than someone Kramering into a space trying “to help” without spending any time or effort trying to understand that space.

They should have understood the guidelines before turning on their machine translation in a given locality.

It's also entirely possible that the Japanese translation team didn't put the guidelines in the right place for the bot to follow them. Or there's a bug. Since another commenter says it's following guidelines just fine.

> I would consider it outright insulting if someone who ostensibly "wants to help" doesn't know basic information like that

Well, the person who wants to help is a customer service manager in Indonesia. They presumably are not the leader of the machine translation product. They are trying to get more information so they can, you know, escalate to the right people.

So how is that "outright insulting"?

Turning off the machine translation and reverting all the changes it made seems pretty actionable to me. They can turn it back on when issues are addressed.

Indeed. Turning it off would still satisfy marsf's requests:

- I prohibit to use all my translation as learning data for SUMO bot and AIs.

- I request to remove all my translation from learned data of SUMO AIs.

Before fixing it and re-enabling it in some capacity, they could work with marsf to find a solution.

They are actionable by entirely canceling the machine translation operations in that community,

even if that were the case (others have explained why that’s not so), that would be an inappropriate time to apologize. you don’t apologize for how someone else feels. you apologize when you recognize that you did something harmful and when the harmed party is amenable to receiving it. otherwise, you’re really just being a jerk who’s only acknowledging that you don’t like how someone else feels.

This isn’t out of the blue. The responder was the lead on rolling out this new translation workflow to the community. Check out the thread (https://support.mozilla.org/en-US/forums/contributors/717387...) and you’ll see people immediately called out the exact potential problems the OP here is complaining about, nearly three months ago. This is no need for the responder to follow up to better understand the complaints. They know exactly what the issues are.

The third, more likely option is that it was a careless act. Clearly a mistake in any case.

What's the saying?

> Never attribute to malice that which is adequately explained by stupidity.

At some point the rate of stupidity becomes suspicious again.

Then say so if someone complains about it? "Shoot, we accidentally ran it in the wrong mode, sorry about that, I'll look into someone fixing it ASAP" is a lot better than "We're sorry for how you feel".

By one side, we all know[1] that automated translations are always inaccurate, and this is specially deep with translations into languages like Japanese.

On the other hand, we all know that LLMs are guaranteed to introduce an error in the result, therefore always needs human supervision.

Also there are other details like the 72 hours timeframe they comment...

Therefore, to me, introducing LLMs without designing them just as an usable tool (UI/UX) to support the work -of the volunteer community- doesn't sound like a mistake, but sounds like an "I don't care at all, get out of the way".

[1] who speaks more than one language at least.

The damage is done when the mistake was allowed to happen.