I've done a bit of reading over this thread. Okay, well admittedly, skipped over many posts and some altogether, as I've only so much time in my day and figured I'd spent enough time reading this thread already. I also know that you're all on the edge of your seats waiting for this reply, so again I'm probably for the most part wasting my time (and potentially yours by reading another potentially meaningless, opinionated post). Still, I feel compelled that I must say something (that's what happens often when you read other opinions, reflections and information, you suddenly can potentially be overcome with a whimsical -or more serious- feeling of needing to put yourself into the mix). I might add however that given how little time I have, I'm eating into my Norwegian study time, French reading time and so, my promise to myself is to not respond to any replies to this post nor to get involved in any more discussions before getting my languages in order (okay, that's a little impossible, but you probably get the point)....
Le Baron wrote:Might as well retire the human race and just leave the AI to it. We'll see if they're any better at politics.
To be honest the excitement around this topic is very tiring now. It's such a one-dimensional view of 'learning'. The deep core of language learning is interaction and the communication/understanding of human feelings, ideas and cultural experience. Not simply 'learning 'facts' and being 'efficient' and getting corrections. There's some of that, especially at the beginning, then it starts fading away.
The faith that somehow the better the AI, the better people will always learn is a curious faith. Any tool is only as good as the user and in this case, because the AI is powerful enough to do something doesn't mean this magically transfers to the user's capacities or benefit! The advent of the internet and smartphone with information and learning tools everyone's fingertips doesn't seem to have made the greater mass of people particularly smarter, perhaps even more stupid. Thinking effort and memory has already been partly abandoned in favour of search engines and cutting and pasting from Wikipedia (or maybe that was already there and was easily transferred). I always thought it was about methods for improving human interaction and understanding.
Teacher's should, as usual, be worried about one major thing: 'service providers' eyeing up AI to eliminate their employment costs and individual students being misled into doing the same.
Mark me down as a naysayer.
You need to read Technopoloy
by Neil Postman. Written 30+ years ago, never more relevant. I'm after finishing it myself a few days ago. I also agree about what you said about learning and would go so far to argue they don't learn
. I really wonder how much input they require, and how much input that is compared to a human that 'learns' to talk; I'd wager the human is doing a lot more with a lot less (like, actually understanding semantics, idioms, etc and not giving back hallucinations), which shows a huge fundamental difference between LLM 'learning' and human learning.
That said, I'm a naysayer myself. Why? Simply because ChatGPT can't detect what is actually correct
. This matters a lot for languages like Irish, where most Irish is you see online, and even most Irish found in EU documents, etc., has lots of mistakes and is simply wrong. Sadly, I've already seen people trying to build apps to teach Irish via AI, and it's set to be a mess. Garbage in, garbage out.
Disclaimer: I am very
against AI chatbots in general and think they don't solve any worthy problems and will actually just make humanity worse off
in pretty much every way. I hate this whole idea of pushing forward with it without regard to societal consequences that everyone seems to have currently.
Monty wrote:When AI takes away the jobs of 90% of the people (including 90% of the members of this forum), and AI is used everywhere for surveillance and repression, we'll see how much enthusiasm is left for it.
Don't have a job? Retired already? Good luck talking to a chatbot trying to fix an error in your state pension. Good luck with "I'd like to talk to your supervisor".
At that time the standard refrain will be "But we didn't know!"
Yes, you did.
Mark me down as a naysayer as well. I don't hold particularly positive views of AI, generally because I don't trust things 'sold' to humanity on a large scale. It's a bit like mass media hype on whatever topic. Why the hype? Well, there's an objective, one which hopes to see humans en masse behave in a particular manner. I think AI potentially has negative goals, but it may not at all, just my opinion, and an opinion I'm not 100% on, as let's face it, I'm no psychic (so don't anyone accuse me of masquerading as one!).
I think, however, having read a lot of the posts now (yes I did read many), my views on AI were incorrect somewhat. Let's be clear, I'm not IT nor AI expert, but let's also be clear, I do feel I have a good sense of when things are being pushed on humanity often for nefarious reasons, but of course that's personal opinion. Nevertheless, I was more afraid of an artificial intelligence actually
learning for itself, much in the way a human could, but
, no, faster. It doesn't seem that AI actually does this (and I'm regurgitating what I've read from others' posts), but rather takes a mass of samples and 'learns' from them in a way that it reproduces simlar-ish 'things' following the high volume of samples it has 'digested'. If AI stays this way, I don't see much of a danger in it in the way that it could threaten humanity's existence, and while my 'Danger! Danger!' signals are dimmed and I feel a sense of surprise (that I was wrong, pfft, who would've thought, given I am the IT expert that I am), and relief, I do still feel we don't really need this. No surprise though, people will create, and people have created this.
Irena wrote:Not legal? Good! And good to know the Irish haven't quite lost their marbles.
Heaven forbid some people think saving a language and culture is more important than letting a bunch of English speakers move wherever they damn well please, as if they're entitled to be able to go anywhere and have everyone cater to them without giving a damn about the locality. It's the same mindset that exists behind AirBnB, which is destroying communities everywhere
. Also, housing discrimination is common, under several names. Planning and zoning is one name for it.
You want to restrict where people can live based on what language they speak. If that's your cure, then it's worse than the disease. AirBNB, planning, and zoning have nothing to do with it.
Changing topic a bit... You know this is what they do in some countries (how many IDK) already. Many countries (okay I at least know of one - the Netherlands) have a certain level expected (of the national language) by a certain date or you risk... umm... something?
I actually don't know the consequences, but I know that in the Netherlands, when I was there in 2011 it was policy that your level of Dutch had to reach level ___ by a certain date, proved through an exam. I'm almost certain other European countries have similar policies. Still, this policiy as far as I know, doesn't apply if you come from other parts of the EU, which I think is stupid. If you are trying to culturally and linguistically protect your, well culture and language, then why should free movement take precedence? All in all I agree with such policies. It's not like you're sending people to labour camps. In fact, it arguable aids in 'participation' and diminishes (somewhat?) the risk of fracturing societies in to sub-groups more independent of each other than would be the case without such policies.
Cainntear wrote:AIs have to be deliberately programmed and/or trained to avoid picking up views that are considered objectionable.
Yes this is one of my concerns. Who controls what's considered objectionable? Eg. If extreme leftest views are deemed as favourable, then it's more likely that AI will be programmed to 'think' in such ways. Is this fair? Or if you
(that's 'you' as in the reader, not necessarily Caintear) prefer left-wing views and live in a country/area governened by those with similar views, how would you feel about AI having been programmed with right-wing views? AI risks teaching others how to think. Reminds me of something.... mass media! More propaganda! I mean, sorry, more 'correct' views.
tastyonions wrote:For Cainntear and galaxyrocker: what's the realistic or even feasible plan for maintaining actual Irish (not anglicized zombie "Irish") as a living language?
Along with what Cainntear said, we need to drop the whole idea that schools are going to save the language. This idea is very
prevalent in Ireland, and has been since the founding of the state. They thought - wrongly - that the National Schools were the reason for the decline and therefore the way to improve it.
My plan would be to focus on the Gaeltacht, and only the Gaeltacht. Drop the mandatory Irish requirement nationwide, first and foremost. Yes, there'll be fewer people taking it, but those few can have better
teachers (most teachers are, honestly, quite inept and can't even distinguish /k/ from /x/, let alone palatalised and velarised consonants!) and actually focus on learning the language as it's spoken in the Gaeltacht.
Then, subsidise people from the Gaeltacht becoming teachers, and place them at schools in and around the Gaeltacht
. Start with the breac- and iar-Ghaeltachtaí, the places that were Irish speaking within living memory and work outwards. All Irish schools there, taught by native speakers from the area, with the dialect, so they can connect to the local place. And then slowly expand. Also start including grants for more and more business in the Gaeltacht working through Irish
. Make it mandatory, and if the business doesn't use Irish, it pays a fine. Also, housing discrimination to stop the gentrification of these areas.
It's not quick, it's not flashy -- and it's probably not legal -- but it's basically the only way.
I'm not an Irish learner. I've not experience on the matter but I appreciate your reply
Back to Norwegian.... if I can find the time (and not distract myself)....