Quite an arrogant answer, as I never said this came as a surprise. I’ve swallowed quite a lot of AI-generated content before bringing my opinion on this forward, so I think the attitude is unnecessary. I’ve built my career in the IT sector, so I know very well why the change has been made the way it has. I still consider this a rather clumsy underestimation of the customer.
Nor would I appeal to the fact that some other garbage forum has suffered more from AI content. I think that’s a really weak argument. It’s purely a choice of the administration regarding how badly the discussion culture is allowed to erode with technological development. It is also an ethical choice to allow ecologically unsustainable garbage to be spewed into the ether without a reasonable cause.
Thanks @Musti for the comments! The intention is by no means to hide behind Discourse, but to mention the limitations coming from there, which we are solving with them. There’s no getting around the fact that the forum is built on the Discourse platform, which limits our ability to create customized solutions. We are trying to solve things smartly regarding the platform’s technical foundation vs. making our own makeshift solutions that break during updates.
Hopefully, we haven’t stated anywhere that this would be a finished feature We are developing the service with a relatively small team together with users, and that approach includes iterating with beta versions. Apologies that this particular feature has now caused trouble!
Thanks for going through these! It includes at least broken messages (token limit on Discourse’s side, which should be resolved with the next update), wondering about the translation order of messages (the queue is stuck and its clearing still needs to be optimized), and misunderstanding words (the persona and prompt need to be tweaked). As you can see, in all these raised points, improvement ideas have already been identified, which we intend to try out in the coming weeks. Unfortunately, not everything has been done yet, as we also need to create value for users and Inderes shareholders in other ways
Yes, I understand this very well, and that’s why it’s really good that feedback is given on small and larger issues
Apologies @Maitopoika that you feel this way. The intention was not to be arrogant; I just wanted to state that this has indeed been prepared, communicated, and tested. In terms of the process, I think it was an 8- performance, but the end result seems to be even weaker. We will improve on this
It’s good that you bring this up. I think it’s important that the quality of this discussion culture is discussed both on the forum and within Inderes. So, the feedback has been received, and the intention is to improve quality in the ways mentioned above.
The intention was not to lash out or blame anyone for being lazy! Mostly, customers are served invaluable here, perhaps that’s why this AI theme sparks discussion ;). I understand, and I am patient!
I didn’t read all the new messages. But for some reason, auto-translation had turned on by itself, even though I had previously turned it off. There were no logins or session clearings in between. It would be great if that could truly be turned off permanently so that it never translates messages.
I’ll elaborate on the translator now. Quite often, users quote English text, either news or direct quotes. AI also translates these, even if the text is otherwise in Finnish. Absolutely infuriating. I wondered for a moment why that text looked so clumsy, until I remembered that oh right, AI is here falsifying writings. It would be good if, when it alters part of the text, it would directly state that this is automatically translated. Yes, it is visible from that button, but it is poorly visible on a mobile phone and especially with a dark theme.
Here is a fresh example of AI’s output in this thread
Thanks, I finally found it. With a dark theme and on a mobile phone, I almost couldn’t see that icon (either). Personally, the best option would be to choose message by message whether I want to translate it.
Above, Faron’s thread was used as a good example. I don’t know, there too it was suspected that they were discussing with an AI bot. The messages were so polished that no human writes like that, and it seems some facts went out with the bathwater too.
As far as I understand, the browser translator is also on here. I also got that kind of mess, which went away when I removed the browser translation from the Inderes website.
Nothing should be on. This happened after I pressed that forum’s translation button, hoping it would turn off. It probably started translating Finnish into Finnish, and everything became a mess like that.
Thanks for the heads-up @Paapaa! That should stay on/off, but I’ll look into this further. I agree that the user should have the option to turn it off without it resetting at some point.
Hmm, strange. In your picture, there is indeed some extra browser translator or phone equivalent on, as @Vanerihands pointed out, because the username supremegod has also been translated to ylijumala (overgod).
That’s exactly what happened: when I was testing the forum’s translation function, it activated the browser’s translation function. I had disabled it from translating English and Finnish in the settings, but for Swedish, at some point, I accidentally pressed “translate automatically.” So, when I changed the forum’s language to Swedish, the browser translated all the texts on the page, and I didn’t notice.
Yeah, it’s sometimes a bit eager and forgets what it’s doing, sorry! Luckily, it can correct unfinished ones, and for me, it looked perfectly fine now.
I think this is partly due to our current translation pace, as the interest is to get the history translated as quickly as possible. Once that’s taken care of, we can give the translator better peace of mind when there are only new messages
Almost without exception, they are so poor that no one would dare to call them translations at all in good conscience. This feature is so thoroughly plagued by teething problems that it’s becoming difficult to understand why this grotesque pilot isn’t being pulled back already. Whose authority is at stake here, when an obvious decision, the necessity of which everyone probably sees, cannot be made.
Yes, as @Vanerihands noted above, it is now possible to edit translations of your own messages. Mods can also do this, as @Sijoittaja-alokas noted above.
I have commented on the partial breaking of translations mentioned by @ljkangas above in previous messages. Yesterday’s Discourse update should resolve this and a few other translation bugs. The platform or forum is not yet complete, so there is still room for development and improvement. Thank you for your patience!
Yeah, this is strange, I’ve tried to force the title translation before but for some reason it gives an error Luckily, I haven’t seen other similar ones myself.
We are developing the service with a relatively small team together with users, and that approach includes iterating with beta versions. Apologies that this particular feature has now caused trouble! 
Luckily, I haven’t seen other similar ones myself.