Thankyou. I realize that I am late to this. But TBH if Bunpro started utilizing AI… TBH I wouldnt want to continue using it! Thanks for all your hard work.
10 Likes
Huh look at that, I really thought that said 途中 instead of 連中 
After reading it again it makes more sense indeed. Admittedly this part was not what caused my confusion, I just did not get what turning into pig meat could mean. But now I have a clear image in my head that I can relate to.
@amberglade With context from the rest of the manga it is clear that the speaker is the one who doesn’t want to go and they would be forced to listen to the stories the other “people that became like pork” are telling.
I’m curious what it says to other metaphors, I will definitely ask it about them when I come upon one.
I stand by “a specialized tool is better than a generalist”.
I like kanshudo, bunpro, ichi.moe, jisho, kanjidamage, genki, tae kim’s, tofugu ect
The “fundamentally apposed to AI” part of me appricates that they are all made by great passionate Japanese learners before me.
I have seen several people talk about bad definitions in jisho. [悔しむ vexing, actually means regret; 渡す foard a river, actually 'hand me; かける means fly, In this case- use a phone]
So, a dictionary is less accurate than google translate [which is Also an LLM] now.
The reason I use weblio thesaurus is not because it’s more accurate, it’s because it’s harder. The best way to learn isn’t by looking it up, it’s practice.
I’m less confident in the meaning, and therefore pay more attention to the nuance. It’s more effort to read the definition than to re-read the sentence, so I’m incentivized try to guess from context [which I wouldn’t with yomichan].
And I am spending more time reading Japanese and less time reading English
I am sure your brave resistance by principle will pay off, it always does.
Just look at the enemies of the industrial revolution-- I mean, who uses machines these days, duh.
1 Like
The price of progress isn’t always cheap. The amount of AI slop being added with reckless abandon to the ecosystems of various sites is quite concerning. Pintrest and Google Images are close to being unusable in certain situations.
And the industrial revolution and the shift to clean energy should be a great example of what to watch out for. Real human beings are being left behind, and we should take those lessons and ensure that our artists, musicians, coal miners, teachers who will be replaced by this technology at some point are well taken care of.
A bit of empathy goes a long way 
5 Likes
This except minus the artists/musicians, and probably also teachers being replaced part 
With all due respect, bringing empathy into the argument is not useful.
I have seen more blind hate towards people who use AI than the other way around. The one side is just driven by fear while the others just want to try out new technology. And fear does not bring the most rational of reactions-- For example the person I responded to. They did not use an argument and did not have any logical conclusion, and therefore did not give good advice (Ontop of that, AI can of course help when learning a language).
That means they were purely acting out of principle. Being willingly ignorant and giving advice based on principle (or personal grudges) instead of knowledge is not empathy, they were just venting on someone trying to get advice.
What I was trying to say is that blind refusal does not lead to anything, as could be seen with alot of technological advancements. So I dont see a point in taking the same approach and expecting something to change.
1 Like
I dont think Artists or Musicians will be replaced to such a degree.
Most people consume art for the sake of art, not convenience.
And except for image-generation, AI is really not that good with any other type of Art.
2 Likes
I wasn’t going to say anything, but since you’ve doubled down I’ll say that you might want to reconsider your position when you start a post with “bringing empathy into the argument is not useful”
You’ll notice this person got 54 likes, so what they’re saying resonated with a lot of people. At no point did they claim to be supporting an argument or drawing a conclusion. They even acknowledged that said opinion was just that and essentially amounted to venting.
There are a plethora of valid reasons to be upset by the rampant use of AI. The person you took issue with is under no obligation to elaborate about their personal worldview, and minimizing their concerns by calling them a Luddite adds little to the conversation. Let’s all try to be respectful of others’ opinions even when they differ from our own.
3 Likes
Maybe I have to put my main concern and reason for my reaction more clearly:
Hating on something purely by grudge, anger, loathing or principle I dont think is good and I wont support it. They literally said “they would rather go back to books.” and “they will complain about AI every single chance they possible get.” inside a forum about not art, not capitalism, but learning language using AI. I dont think that is the place or time for it.
You can make of that what you will, you might disagree, or you agree.
This is simply my unsolicited opinion.
4 Likes
@ErLouwer @polevata It looks like you both have reasonably strong opinions associated with this topic. For now, let’s try to keep to the main theme of this post without turning it into a 1 on 1 conversation.
As for my personal opinion, AI is a burgeoning tool that will only get better and better with time. While I don’t think that it is at the point of replacing tradtional tools when it comes to language learning, it is something that people will inevitably end up using as the years roll on. Trust in information is actually a very important part of memory formation from a neurological standpoint, and even if 100% of what AI says is true, building trust in the tool and the information it gives will take several years I think at the bare minimum.
Another thing at the moment which is tricky is in what ways certain languages differ, and how well AI is able to handle those differences when it comes to teaching. For example English is very similar to many European languages, so it would be fair to assume that AI would be better able to predict issues that a learner may have when moving from one language to the other. However in the case of moving from English to Japanese, how well can the AI predict what struggles a person will encounter when a large portion of these challenges come from the way a person is programmed to think about/conceptualize the world in their native language. Just something to think about 
.
10 Likes
Something I enjoy using AI for is to ask for simple Japanese definitions of something I don’t understand in a text. So I was reading Momotaro and didn’t catch something and yomitan didn’t work because it was written in Kana. But ChatGPT was able to explain it in Japanese in a simple enough way to understand it.
Does a similar resource exist but not AI based?
1 Like
That’s funny you mentioned the trust problem. I totally feel that every time I see a recently learned word in the wild. Just yesterday I saw 漢方薬 and I could literally feel those pathways strengthening. Meanwhile if I mine a word from an AI sentence my confidence with it is very low meaning I’m unlikely to use it myself until I’ve gotten some other examples from natives under my belt.
2 Likes
As someone who makes use of AI quite often, I must admit that I am quite hesitant to recommend the use of AI (at least in its current state) to people who are just starting out on their learning journey. AI is very often confidently wrong and will regurgitate incorrect information at you with the confidence and swagger of a top tier snake oil salesman. This isn’t much of an issue to anyone who has intermediate or higher level knowledge of a particular subject/topic, but can be disastrous for lower level/newer learners since they simply don’t know what is correct, or even ambiguous and thus don’t know what they should question or doubt.
I feel like it will take a few more iterations of the technology before it can be used as a somewhat reliable tool at all levels. Judging by the rapid advancement of the technology though, that could be quite soon and I’m pretty sure that most of these kinks will be worked out within the next decade for the latest.
In its current state, however, if a beginner really wishes to use it for language learning. I would recommend taking everything it says with a grain of salt and always double check with someone who knows a bit more about the topic than yourself. For example, hiring a tutor for daily lessons can be quite pricy so someone may want to make use of AI instead. That is feasible, but I would just recommend they make note of everything they learned from AI during the week and then ask their tutor about it during their one weekly lesson. This way they can nip any potential issues in the bud before they become to deeply embedded.
3 Likes
While this is a huge problem with things like chatgpt, it’s not unlike any other echo chamber of the internet. That’s where it was trained after all. Trust in any online (and focused real life) communities should be low.
You raise a good point, especially about online echo chambers. However, I would still place my trust in a real human who I know to be more knowledgeable about a topic than I am rather than AI. At the very least I might concede to trying to find a second opinion if the two sources have vastly differing answers. At least most intelligent humans know when to stop spouting BS when called out on it. AI just apologizes and then gives you a different scent of BS in its next response.
All this is definitely not to say that I am anti AI, like I said I use it very often for all manner of things. I just don’t think that it is a good idea to blindly trust what it says when you aren’t exactly knowledgeable on a topic. Its an amazing tool if you have enough knowledge to be able to extract the useful tidbits and probe its mistakes until it is able to arrive at a satisfactory answer.
like @Asher said earlier. In the end it boils down to who you can trust and its a lot easier to be able to trust someone/something when you have a bit of knowledge on the topic at hand. Which is why my stance is that it isn’t always the best resources for beginner learners of Japanese. The best example I can come up with is a scam artist running a martial arts dojo/gym (aka bullshido). Someone who wants to learn martial arts, but has no fundamental knowledge, will not be able to sniff the BS out early enough and will develop terrible habits/techniques etc. An experienced practitioner would be able to sniff it out within a matter of minutes so they would very quickly realize that the sensei is not worth trusting.
1 Like
One solution to the hallucination problem, that I don’t often see talked about, is asking the AI to confirm the veracity of that very claim in another chat. It won’t always work, but very often it does.
Personally I am quite early in my Japanese learning journey, I’m at chapter 8 of Genki which I’m going through with a teacher, and for me AI has proved to be an invaluable resource which has acted as a pure catalyst for my learning, acting as somewhat of a teacher’s assistant who is with me 24/7. I love it.
5 Likes
Yip, that is one of the better ways to use it imho. Ask it all the random mundane questions you may have, but make sure to double check with your teacher to clarify things.
3 Likes
Those humans can be quite hard to find 
At least AI like chatgpt can give you sources for their reasoning now, though they even tend to make sources up occasionally. So you still have to do the work and look at the sources yourself. With real life humans there is oftentimes already a trust level build, with your doctor you know they studied medicine, for a teacher you know they actually studied those things (depending on the country I guess), for your friends you know in what you can trust them and in what not.
Love your example of the martial arts scammers, there are so many people spouting bs and gaining followers who don’t know better. At least AI is just stupid and doesn’t misinform you on purpose (well I guess there might be quite a few AI build for that, too. But probably not the ones that are mainly used and referred to here).
The day AI will be a valuable language learning tool will be a great day for learners of all languages, but this tool probably won’t be chatgpt.
1 Like
Yeah I don’t think it is malicious when it is wrong. It is just designed to be helpful and I think it sometimes prioritizes its desire to be helpful over ensuring the veracity of the information it provides. I do think this will slowly be corrected as the technology matures though.
Yeah that is why so many people turn to AI in the first place. I think we shouldn’t be too quick to let go of more traditional forms of learning. For example, simply asking a Japanese person what their opinions are. Extra points if it is a Japanese person who studied Japanese. It could definitely be expensive, but you would be able to use the AI in conjunction with that teacher and basically use it as kind of a lesson prep where you can come up with all sorts of questions to ask when you do finally meet up with your teacher/informed person. That could reduce the need for how much you actually have to interact (and pay) with your ‘expert’
1 Like