My first language is Afrikaans, which is a somewhat niche language and hard to find teachers/conversation buddies outside South Africa. (I live in USA now)
I've been using Gemini to live chat in Afrikaans and do impromptu Afrikaans grammar lessons during my solo drives around town. It is phenomenal at speaking the language - like, it really shocks my family members when they hear it.
This is probably the most joy I get from any of my usages of LLMs/AIs. It's been really, really nice getting to speak my language regularly again. =)
So, I'm excited about this release and live chat getting better. I also hope the other frontier labs pick up niche languages like this as well so that I have more options.
Yeah Gemini has been consistently better than open ai's chat in icelandic, but I would still say it's far from passable as natural sounding. Lots of grammar errors and the pronunciation sounds like a non native speaker.
for me the live mode in androids google translate app has been as close as it gets to perfect for traveling cannot believe it is a free service after trying so many others
I have a similar experience using Gemini for quick Catalan translations for iOS apps given enough context.
I once asked it to summarize The Hobbit in Catalan to explain it to my daughter before sleep. I was expecting a lot of mistakes as I see regularly if I ask anything in my native language when using GPT or Claude, but it was surprisingly good. I was going just to kind of skim ahead and retell it my own way, but ended up almost saying it verbatim because it was good already.
She loves Zelda so I asked it to explain the story of Breath of The Wild keeping the original names, and to make it fun, etc.. I was surprised again. I did retell some bits in my own style and taste but it is very convincing.
I haven't tried Catalan on newer models like GTP-6 Astra or Fable tho. We have all these benchmarks based on software development, and AGI, etc.. but it would be cool to have some language benchmarks for different communities.
As I work in english and use them in english, I wonder if using LLMs in a different language to code renders a different result as well. Like, if some of these benchmarks were made in other languages, would the result be similar.
> As I work in english and use them in english, I wonder if using LLMs in a different language to code renders a different result as well. Like, if some of these benchmarks were made in other languages, would the result be similar.
I am not a Chinese speaker but my understanding is that all of the models have substantially different behavior in Chinese, to the degree that it's kind of like a second model. Would be interested in hearing more if anyone has direct experience.
Fwiw, it's been very easy to nudge Qwen 27B/35B to think in Chinese. No need for <think> prefills like "思考：" ("Think:") or such.
Relatedly, when chatting with frontier models about science education content design, I've found it very helpful to mix in Chinese education terms. In English, for example, NGSS is such a massive attractor, discussing nearby topics often yields NGSS "slop". And "estimation" (educational) in the US means one (dysfunctional:) thing, which similarly distracts. Perhaps if AIs become increasingly multilingual, but remain weak at deep conceptual reasoning, it may be fruitful to have multilingual thesauruses, to use language-associated cultural conceptual differences as a way to convey conceptual nuances with which LLMs otherwise struggle?
This is an interesting post, but as a non-American it's hard to understand what's it's alluding to in some parts.
The one criticism of NGSS I found from a quick skim of <a href="https://en.wikipedia.org/wiki/Next_Generation_Science_Standards" rel="nofollow">https://en.wikipedia.org/wiki/Next_Generation_Science_Standa... is that it teaches evolution.
> What does "estimation" mean in the US?!
"Estimation" as a skill in US education, primary school on, is pervasively about point estimates - single numbers. Bounding, a range of safe/reasonable/likely/possible numbers, is very rare. In Fermi questions/problems also.
Even in such goodness as Sanjoy Mahajan's Art of Insight and Street-Fighting[1]... the only only "bound" mentioned is "Printed and bound in [USA]", and there is no "round".[2]
In contrast, IIUC, China is pervasively about "Da Gu" and "Xiao Gu", "round up" and "round down". Point estimates are secondary. Lower and upper bounds are apparently even a cultural concept: an individual's degree vs capability/fate; policy equity vs excellence; economic targets.
In education, my understanding is point estimates have much poorer conversational/collaborative dynamics than range estimates. Bounding lends itself to incremental collaboration. "Can anyone suggest another low or high bound?". Discussions around point estimates seem usually merely about the collection of individual estimates.
> NGSS
Ah, NGSS isn't bad particularly. I meant that discussion/mention of NGSS appears so very much in English training data, that AI chat on topics related to NGSS seem to draw regurgitations of NGSS. AI "slop" in the sense that "ignore NGSS, just reason from a superset of its underlying concepts"... isn't an LLM strength.
ERRATA: Those google book links now do have hits for "bound" and "round". Also "range" and "interval". The searches really didn't before, thus the reference links. Perhaps the absence should have failed sanity check, but given that one hit... partial failure of Google Books search wasn't something I recall noting previously. Oops.
Yes, and that brings in a shit load of slop to the problem space from the creationists.
Like how the word life will have a strong correlation with death, the word animate, while it also means alive will have a much less strong correlation with death.
Does European public dialog do much causal comparisons among states? Does somewhere else?
Or do you mean professionally... Instead of merely searching for "ways to do X", one might translate the query into a couple of languages, and translate the results, to get a broader culture sampling. Hmm... That seems something an AI-based search could integrate smoothly - cross cultural insights as a normal part of search results?
While there's a lot of diversity in policy and culture among US regions and states, contrasting them isn't leveraged much. Perhaps failing a "that's analysis, not news" threshold for being part of public dialog? So maybe it's not "non-English cultures" neglected as much as comparative discussion in general?
One "criminally underutilized" which really struck me, was Japan having a far more sensible and powerful COVID safety one-liner than the US managed - "Three C's" (Closed spaces, Crowded places, and Close-range contact - poorly-ventilated / crowd / conversation-contact - avoid overlapping all three) vs "6-foot rule".
I love Gemini in Google Maps for long drives. I start getting bored of music and podcasts and start grilling it with random questions I've always wondered about.
I do this too! Usually some idea in ML ...when I am driving I might suddenly remember what I was thinking about and then it is my personal podcast via Gemini-in-Maps. My only complaint is if you have follow up questions, you have to be quick, otherwise it cuts off the mic.
Even as a human being who has a working brain and all this thinking power, I am not sure how I would answer that. It's kind of funny Gemini will still come up with something and even suggest follow-ons to the conversation but saying all that stuff as a human to another human would be a wild response to that question.
Really? I am in Android Auto and I try to even just ask it to get me to "Costco Near Me". I know where it is. I wanna know if the sometimes there traffic jam that blocks the right lane is gonna expect me or not without being distracted zooming out from current location and then back in to the highway exit I know is prone to that at times. You know, I wanna be a good citizen and not be distracted by looking even a second at the screen I'm zooming around in.
It tells me there's a Costco 10km from me and I'm like, yeah let's go there, that's the one I meant! And then it sends me to one that's like 40 minutes from me all across town and I'm like WTF?!
Never mind asking Android Auto any question that's not driving related. It just doesn't understand me at all. Or even some driving related ones. "Find Alternative route". "Sorry, I can't help you with that". WTF?
Does you Android Auto show a microphone for voice our the Gemini star? I think you're likely talking to the pre-Gemini voice assistant which is very basic.
Gemini Live now can be used to talk to any article on the web. I asked questions on the Gemini Live launch, got examples and use cases all live. <a href="https://app.digitalu.live/demo/6FRMzTD-pv" rel="nofollow">https://app.digitalu.live/demo/6FRMzTD-pv
I've been building a language learning app which uses AI voice chat to let people practice. Would that be something interesting for you to try out? It's all still rough, but we've been learning Japanese with it and its pretty cool.
We (me and my teens) used Duolingo for a few years and.. didn't really get that far in actually being able to speak. It's kind of mind-blowing that they are the market leaders. In retrospect, I think of it more like an addictive language game and not really a good way to actually learn to speak.
If you're really interested, shoot me an email.
Latest in a long line of scams for learning a language. Goes back to cassette tapes and records. The only good way is the natural way: immerse yourself in it.
While in general I do agree with you (it's how I learned Italian, Hebrew, and Thai) being on the ground for a year is cost-prohibitive for me right now, and unreasonable for kids under 16.
What I've been building, with the help of a few language teachers, is our best guess at a really good way of learning new languages from the comfort of your home. It's not ever going to be perfect, but it is pretty good (at least so far in testing within our small group), and very cost-effective.
If you're interested in funding it, I'm happy to go live in Japan for a year :)
My father-in-law was saying the same thing about Gemini 12 months ago regarding the Afrikaans speech. I've tried a few of them in my studies but none of them could seem to switch between English/Afrikaans except Gemini. It's an interesting time to be a language learner
As jy wil, ons kan saam praat op Discord :) maar my Afrikaans is sleg
As a Dutch person i find Afrikaans always interesting whenever i happen to encounter it in the wild. As if a common ancestor took a very different evolutionary path.
It's fascinating to look up the etymology of words and see how it looks in Dutch. I find Afrikaans to be much closer to English than Dutch or German (which I previously learned). Do you find Afrikaans comprehensible?
Did this with shona.On the road trip we (3 human occupants) spoke to Gemini in ndebele, karanga, portuguese and ndau and have switch seamlessly to another lanuage like german without loosing the thread of the conversation. what impressed me was its ability to recognize the subtle dialects of shona being spoken, articulate its unncertainty, ask for clarification and carry on. when asked to speak those dialects it politely lamented it could not. when pushed it insisted and basically apologized in a technically correct but awefully pronounced sentence of the dialect before going back to english.
Hey I literally just did the same thing after seeing the Afrikaans reference ndikawona kuti zvirikushanda! I was incredibly surprised, its shona was really formal but that makes sense when thinking about getting training data.
I like it. I had a friend at school who came from SA and spoke Afrikaans - so I know how a natural sounds like.
For language learning, translation, research - it is fantastic.
It works very well for certain dialects as part of a local language heritage as well: From Welsh to Bavarian - it is a joy to really get approval from people used to these dialects as being really accurate.
Lekker man. Although I speak it often, I could use an LLM to help me progress as I've been stuck at my current level for about a year and typing it is not something I enjoy.
I find no greater joy than listening to a TTS model say Poes (Google translate is one of the funnier ones to make swear). Reminds me of being 14 lol.
I am assuming these languages are present in translate.google.com If Google Gemini is good in languages other than English, while other models are not, then we can trace back it back to the translate.google.com as a resource as one possibility.
I can't wait to try this! Gemini was already great at languages and just general spoken conversations.
Me and my wife have been going through a rough time of it with some health issues, and talking to Gemini has been a boon through this period. Both of us are Indian, and in just the past day we've:
- had insightful conversations on the diagnosis challenges of complex health issues that cross the mind-body connection that have helped us have more productive conversations with healthcare providers in the US
- had fun conversations switching from hindi (spoken in a punjabi accent) to pure tamil to "city" tamil and then back to english (this time in a fake aussie accent).
- used it to recommend some furniture we're looking for
- come up with a specific south indian recipe with unique constraints (no oil, no dairy, only ingredients x, y and z in the fridge)
I had a funny moment with ChatGPT. I was talking to it in English and then asked it to speak Swedish with me. It did, but with an English accent. But if you just start a session in Swedish, it speaks Swedish perfectly with local accent ( I guess Stockholm accent is closer to the truth)! Just thought that was really interesting.
jeanbza · · focus · HN ↗
I've been using Gemini to live chat in Afrikaans and do impromptu Afrikaans grammar lessons during my solo drives around town. It is phenomenal at speaking the language - like, it really shocks my family members when they hear it.
This is probably the most joy I get from any of my usages of LLMs/AIs. It's been really, really nice getting to speak my language regularly again. =)
So, I'm excited about this release and live chat getting better. I also hope the other frontier labs pick up niche languages like this as well so that I have more options.
arnorhs · · focus · HN ↗
trollbridge · · focus · HN ↗
jimmySixDOF · · focus · HN ↗
LluisGerard · · focus · HN ↗
I once asked it to summarize The Hobbit in Catalan to explain it to my daughter before sleep. I was expecting a lot of mistakes as I see regularly if I ask anything in my native language when using GPT or Claude, but it was surprisingly good. I was going just to kind of skim ahead and retell it my own way, but ended up almost saying it verbatim because it was good already.
She loves Zelda so I asked it to explain the story of Breath of The Wild keeping the original names, and to make it fun, etc.. I was surprised again. I did retell some bits in my own style and taste but it is very convincing.
I haven't tried Catalan on newer models like GTP-6 Astra or Fable tho. We have all these benchmarks based on software development, and AGI, etc.. but it would be cool to have some language benchmarks for different communities.
As I work in english and use them in english, I wonder if using LLMs in a different language to code renders a different result as well. Like, if some of these benchmarks were made in other languages, would the result be similar.
jaggederest · · focus · HN ↗
I am not a Chinese speaker but my understanding is that all of the models have substantially different behavior in Chinese, to the degree that it's kind of like a second model. Would be interested in hearing more if anyone has direct experience.
mncharity · · focus · HN ↗
Relatedly, when chatting with frontier models about science education content design, I've found it very helpful to mix in Chinese education terms. In English, for example, NGSS is such a massive attractor, discussing nearby topics often yields NGSS "slop". And "estimation" (educational) in the US means one (dysfunctional:) thing, which similarly distracts. Perhaps if AIs become increasingly multilingual, but remain weak at deep conceptual reasoning, it may be fruitful to have multilingual thesauruses, to use language-associated cultural conceptual differences as a way to convey conceptual nuances with which LLMs otherwise struggle?
eru · · focus · HN ↗
For that particular attractor, even German or Spanish or so might help avoid it? Or perhaps even just using British English?
mncharity · · focus · HN ↗
normie3000 · · focus · HN ↗
The one criticism of NGSS I found from a quick skim of <a href="https://en.wikipedia.org/wiki/Next_Generation_Science_Standards" rel="nofollow">https://en.wikipedia.org/wiki/Next_Generation_Science_Standa... is that it teaches evolution.
What does "estimation" mean in the US?!
mncharity · · focus · HN ↗
"Estimation" as a skill in US education, primary school on, is pervasively about point estimates - single numbers. Bounding, a range of safe/reasonable/likely/possible numbers, is very rare. In Fermi questions/problems also.
Even in such goodness as Sanjoy Mahajan's Art of Insight and Street-Fighting[1]... the only only "bound" mentioned is "Printed and bound in [USA]", and there is no "round".[2]
In contrast, IIUC, China is pervasively about "Da Gu" and "Xiao Gu", "round up" and "round down". Point estimates are secondary. Lower and upper bounds are apparently even a cultural concept: an individual's degree vs capability/fate; policy equity vs excellence; economic targets.
In education, my understanding is point estimates have much poorer conversational/collaborative dynamics than range estimates. Bounding lends itself to incremental collaboration. "Can anyone suggest another low or high bound?". Discussions around point estimates seem usually merely about the collection of individual estimates.
> NGSS
Ah, NGSS isn't bad particularly. I meant that discussion/mention of NGSS appears so very much in English training data, that AI chat on topics related to NGSS seem to draw regurgitations of NGSS. AI "slop" in the sense that "ignore NGSS, just reason from a superset of its underlying concepts"... isn't an LLM strength.
> hard to understand
Sorry - Thanks for asking!
[1] open access: <a href="https://direct.mit.edu/books/oa-monograph/5345/The-Art-of-Insight-in-Science-and" rel="nofollow">https://direct.mit.edu/books/oa-monograph/5345/The-Art-of-In... <a href="https://direct.mit.edu/books/oa-monograph/5339/Street-FightingMathematicsThe-Art-of-Educated" rel="nofollow">https://direct.mit.edu/books/oa-monograph/5339/Street-Fighti... [2] search "bound": <a href="https://www.google.com/books/edition/The_Art_of_Insight_in_Science_and_Engine/xRgeBQAAQBAJ?hl=en&gbpv=1&bsq=%20bound" rel="nofollow">https://www.google.com/books/edition/The_Art_of_Insight_in_S... <a href="https://www.google.com/books/edition/Street_Fighting_Mathematics/VrkZN0T0GaUC?hl=en&gbpv=1&bsq=bound" rel="nofollow">https://www.google.com/books/edition/Street_Fighting_Mathema...
mncharity · · focus · HN ↗
pixl97 · · focus · HN ↗
Yes, and that brings in a shit load of slop to the problem space from the creationists.
Like how the word life will have a strong correlation with death, the word animate, while it also means alive will have a much less strong correlation with death.
tomaskafka · · focus · HN ↗
mncharity · · focus · HN ↗
Or do you mean professionally... Instead of merely searching for "ways to do X", one might translate the query into a couple of languages, and translate the results, to get a broader culture sampling. Hmm... That seems something an AI-based search could integrate smoothly - cross cultural insights as a normal part of search results?
While there's a lot of diversity in policy and culture among US regions and states, contrasting them isn't leveraged much. Perhaps failing a "that's analysis, not news" threshold for being part of public dialog? So maybe it's not "non-English cultures" neglected as much as comparative discussion in general?
One "criminally underutilized" which really struck me, was Japan having a far more sensible and powerful COVID safety one-liner than the US managed - "Three C's" (Closed spaces, Crowded places, and Close-range contact - poorly-ventilated / crowd / conversation-contact - avoid overlapping all three) vs "6-foot rule".
asdfman123 · · focus · HN ↗
abhgh · · focus · HN ↗
asdfman123 · · focus · HN ↗
How did the early Roman Empire interact with Greek city states?
How are LLMs planning on learning new information on the fly without new context or retraining runs?
Why did mom leave?
You know, standard stuff.
throwup238 · · focus · HN ↗
Reinforcement learning by human feedback.
Bluestein · · focus · HN ↗
altmanaltman · · focus · HN ↗
Even as a human being who has a working brain and all this thinking power, I am not sure how I would answer that. It's kind of funny Gemini will still come up with something and even suggest follow-ons to the conversation but saying all that stuff as a human to another human would be a wild response to that question.
cm2012 · · focus · HN ↗
jonifico · · focus · HN ↗
rr808 · · focus · HN ↗
tharkun__ · · focus · HN ↗
It tells me there's a Costco 10km from me and I'm like, yeah let's go there, that's the one I meant! And then it sends me to one that's like 40 minutes from me all across town and I'm like WTF?!
Never mind asking Android Auto any question that's not driving related. It just doesn't understand me at all. Or even some driving related ones. "Find Alternative route". "Sorry, I can't help you with that". WTF?
bberrry · · focus · HN ↗
MobzAlex · · focus · HN ↗
HDBaseT · · focus · HN ↗
antonvs · · focus · HN ↗
ramijames · · focus · HN ↗
written-beyond · · focus · HN ↗
ramijames · · focus · HN ↗
If you're really interested, shoot me an email.
SoftTalker · · focus · HN ↗
lIl-IIIl · · focus · HN ↗
You can use it as supplement or do it standalone on your commute before a trip and end up with good pronunciation, vocab and a conversational skills.
ramijames · · focus · HN ↗
What I've been building, with the help of a few language teachers, is our best guess at a really good way of learning new languages from the comfort of your home. It's not ever going to be perfect, but it is pretty good (at least so far in testing within our small group), and very cost-effective.
If you're interested in funding it, I'm happy to go live in Japan for a year :)
maxym_hryniv · · focus · HN ↗
[dead]
geraldwhen · · focus · HN ↗
3stacks · · focus · HN ↗
As jy wil, ons kan saam praat op Discord :) maar my Afrikaans is sleg
FranzFerdiNaN · · focus · HN ↗
aktau · · focus · HN ↗
3stacks · · focus · HN ↗
basch · · focus · HN ↗
It makes total sense it’s good at regurgitating the edge cases of language.
manzout · · focus · HN ↗
zimprop · · focus · HN ↗
_the_inflator · · focus · HN ↗
For language learning, translation, research - it is fantastic.
It works very well for certain dialects as part of a local language heritage as well: From Welsh to Bavarian - it is a joy to really get approval from people used to these dialects as being really accurate.
Happpy times! :)
NorthSouthNorth · · focus · HN ↗
I find no greater joy than listening to a TTS model say Poes (Google translate is one of the funnier ones to make swear). Reminds me of being 14 lol.
orsenthil · · focus · HN ↗
resmofo13 · · focus · HN ↗
brabel · · focus · HN ↗