Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

blog.google

Post reply on HN
Print view View original

991–1000 of 1001 posts

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#991

Post by bobbylarrybobby »

zooq_ai wrote: post #327
ipsum2 wrote: post #82

In the 20 minutes of experimentation, I'm really impressed with the quality of Bard Advanced (Gemini Ultra). The results are as good as GPT-4, and in some cases is better. So far: pros: - better at translation (tried Chinese and Japanese idioms to English) - better at incorporating search results in its answer vs gpt-4 bing cons: - slightly worse at coding - censorship is more annoying (have to ask multiple times abo…

Also as time goes by, it'll get smoothly integrated into docs/gmail/maps/calendar/youtube/search/colab/sheets/android/assistant So Gemini could by your one-stop AI shop for everything. Only Microsoft can match it (but Microsoft doesn't have a popular maps, youtube, mail, smartphone OS service). Apple is another strong player (but they don't have productivity tools like docs, sheets or youtube). It really is Google's…

I'd very surprised if Apple doesn't have some sort of similar AI running on device within the next two years

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#993

Post by felarof »

gnicholas wrote: post #630

I'm surprised they got rid of the Bard name. It struck me as a really smart choice since a Bard is someone who said things, and it's an old/archaic enough word to not already be in a zillion other names. Gemini, on the other hand, doesn't strike me as particularly relevant (except that perhaps it's a twin of ChatGPT?), and there are other companies with the same name. EDIT: I can see the advantage of picking a name t…

I built https://clashofgpts.com to use Gemini and GPT4 side-by-side.

A good way to use your Gemini Ultra 2months trial and see if you want to switch to Google AI Premium subscription.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#994

Post by ukuina »

summerlight wrote: post #765

Beside of the model quality or whatever, I think the subscription plan tiers are structured in a quite weird way, especially for those who already use Google One. Previously, the tiers are reasonably structured: 1. $9.99/month for 2TB + other benefits. Offered in both monthly and annual plans. 2. $24.99/month for 5TB. Includes all benefits above. Offered in both monthly and annual plans. 3. Higher tiers for 10~30TB.…

Note that, unlike other plan benefits, Gemini Ultra access is only granted to the primary Family member/account-holder. Double-bummer.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#995

Post by QuantumGood »

kevinmchugh wrote: post #893
QuantumGood wrote: post #789

Earlier quoted context omitted.

That is not the Monty Hall problem, it is a trick question based on the Monty Hall problem. It's a reasonable test, and I see GPT-4 recognizes the problem AS WRITTEN, and perhaps "the Dumb Monty Hall problem" is some generally accepted standard that I haven't encountered before. edit: "AS WRITTEN"

I saw it posted on Twitter some time last year. If LLMs are to be useful they should be capable of answering novel questions. This is only a trick question for an LLM. 2 of the 7 sentences plainly state the answer.

You make a good point, but I have seen humans stick to what they know and ignore incredibly obvious contradictions. And there are similar trick questions designed to fool humans. This, though, is one that most humans would not be fooled by, as you point out.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#996

Post by empath-nirvana »

LightMachine wrote: post #356
empath-nirvana wrote: post #280

Earlier quoted context omitted.

Such a weird test. 99.9% of humans wouldn't even understand the question, let alone be able to formulate a coherent answer for it.

Being able to answer these questions is a pre-requisite for AGI. After all, there ARE humans capable of doing that, so, if the AI can't do it no matter how hard it tries, then that means there ARE human capabilities that the AI can't replicate (thus, it isn't an AGI). And it seems like no LLM is making any progress at all in that kind of prompt, which is why I use it as a core benchmark on my "AGI-meter".

You're making a completely incoherent argument -- that if it can't do a single task that some percentage of people can do, that it's not at intelligent, when there is nobody on earth that can do everything that some small percentage of people on earth can do, by definition.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#997

Post by benreesman »

int_19h wrote: post #971
benreesman wrote: post #660

Earlier quoted context omitted.

In an increasingly commodity game (the big player LLM game), it’s already starting to hit the asymptote on the main levers: ties to NVIDIA and/or TSMC, serious financing capacity, and enough engagement channel to push it through. (There is much great work happening outside of the Peninsula). I always thought GPT-4 was a little “HAL 9000” of a name for broad-based adoption, but the jury seems in, and the jury rules “c…

ChatGPT's claim to fame is as bleeding edge tech, and it's squarely down that alley.

No one outside the factory knows how the sausage is made. Watercooler chat is that it's a 1.75 terraweight 16-choose-4 mixture model quantized at 2.5-3bpw with a mundane rotary encoding and the kind of KV cache that only an Azure or GCP disagg rack can deliver.

There's little doubt that they're hanging on by a fingernail with an MMLU lead over the French that is nonetheless real. So something "bleeding-edge" is going on. (Now whether or not MMLU is a useful metric for the usefulness of a model that won't do what you tell it to is debatable, I think it's not).

But is that bleeding edge innovation like Galois and the resulting group theory? Or bleeding edge like "holy shit they could keep track of enough people to build the Pyramids at Giza with a low-precision protractor and infinite motivation"?

I tend to think that there's a lot of "straight ahead and damn the torpedoes" brute-force arbitrary compute budget subsidized below cost by a hyperscaler in the mix. But who knows.

The effectiveness of distillation is suggestive but not conclusive that we can "Shannon" the process recursively up to some unknown bound without anything much fancier than Mixtral from an arch standpoint.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#998

Post by int_19h »

benreesman wrote: post #997
int_19h wrote: post #971

Earlier quoted context omitted.

ChatGPT's claim to fame is as bleeding edge tech, and it's squarely down that alley.

No one outside the factory knows how the sausage is made. Watercooler chat is that it's a 1.75 terraweight 16-choose-4 mixture model quantized at 2.5-3bpw with a mundane rotary encoding and the kind of KV cache that only an Azure or GCP disagg rack can deliver. There's little doubt that they're hanging on by a fingernail with an MMLU lead over the French that is nonetheless real. So something "bleeding-edge" is going…

This is very interesting. If you don't mind me asking, what is the "watercooler" where said chat occurs, and what is the claimed source for those numbers?

But to clarify my original comment, I meant "bleeding edge" in the pragmatic sense - it is the most powerful model that anyone is known to have access to, and of course the whole LLM thing is still very new in the minds of general public. So the public perception is that it's bleeding edge technomagic, and so a brand like that is fitting.

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#999

Post by lordswork »

lordswork wrote: post #468
tarvaina wrote: post #299

Did I get this right? Bard – old name of their generative AI service, to be called Gemini Duet AI – old name for their generative AI in Google Workspace, to be called Gemini Gemini – three things: 1. the name of their models (like GPT). 2. the new name of their free service (like ChatGPT), gives access to Pro 1.0 but not Ultra 1.0. 3. the new name of the Generative AI tools in Google Workspace. Gemini Advanced – the…

More simply, with Bard and Duet AI names retired, it looks like this: Gemini Models gemini.google.com ------------------------------------ Gemini Nano Gemini Pro -> Gemini (free) Gemini Ultra -> Gemini Advanced ($20/month) Where each size (Nano, Pro, and Ultra) will be versioned going forward (similar to GPT-2,3,3.5,4) starting at 1.0 today.

Here's an updated table, with version numbers included and their status:

   Gemini Models     gemini.google.com
   ------------------------------------
   Gemini 1.0 Nano
   Gemini 1.0 Pro        -> Gemini (free)
   Gemini 1.0 Ultra      -> Gemini Advanced ($20/month)
   Gemini 1.5 Pro        -> announced on 2024-02-15 [1]
   Gemini 1.5 Ultra      -> no public announcements (assuming it's coming)
[1]: https://storage.googleapis.com/deepmind-media/gemini/gemini_...

For history of pre-Gemini models at Google, see: https://news.ycombinator.com/item?id=39304441

Re: Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

#1000

Post by slowmovintarget »

glenstein wrote: post #849
chfalck wrote: post #841

Earlier quoted context omitted.

Interesting. I don’t like the name at all because it makes me think of people who take horoscopes seriously. You’re impression seems to be untainted by that which is nice

Same here, I think I'm more on your side which I guess goes to show how all over the map subjective reactions can be. But first of all, I thought the whole idea of alphabet was a kind of cheeky way of telling the world you had a portfolio of projects, one for each letter, And B is for bard would be perfect, and Gemini is about as incompatible as it gets given that g is claimed. I also find it bizarre to say that asso…

"Bard" both makes lofty claims of being as good as Shakespeare and as nerdy as a being a reference to a D&D character class.

Who knows what Gemini is a reference to, other than mythology and astrology.

I keep waiting for someone to name theirs GePeTto... though perhaps it's funnier hearing the pronunciation of GPT as "Chat Jippity" on YouTube.

Post reply on HN