Gemini Ultra (we do not have access) outperformed GPT4 in tests performed by Google by few %. I have just tested Gemini Pro - extracting data from medical publications and it performed worse than 3.5. I hope that Ultra will be better - we need some other options than GPT4.
- Google introduced "Gemini," an AI that surpasses GPT-4 in benchmarks. - Gemini is labeled the first Universal AI, excelling in text, code, audio, images, and video. - Three versions of Gemini are unveiled: Ultra for complex tasks, Pro for varied tasks, and Nano for on-device tasks. - Gemini Pro powers Google Bard, offering advanced reasoning, understanding, coding, and planning functionalities. - It outperforms GPT-4 in most benchmarks across text, coding, images, videos, and audio, marking a significant breakthrough in large language models and multimodal AI.
While it's still early days for Gemini, it has the potential to revolutionize the field of AI. It can be used in education, healthcare, the possibilities are endless!
Based on other things they are saying, Ultra is not going to be free. They were mentioning something like "Advanced Bard" and also they particularly stated that pro was made because that's what's scalable enough to put into general web use. So Pro is supposed to be their GPT3.5 while Ultra would probably be subscription based
@@SkillLeapAI I don't think they will. Their statements about having a pro model because that's scalable for general use on the web suggest that Utlra is not scalable for general use on the web and they only have capacity for the few who would pay.
In fact I expect it to have similar bottleneck that GPT-4 is having. Even to the select few who paid for it, GPT 4 is only 40 messages per three hours. What magjc is Ultra gemini using that it's suddenly possible to supply that to everyone who wanted it. I don't think they've made so much progress over the status quo that they can afford to serve something as powerful as GPT-4 to anyone that wants it.
Just prompted Bard with "What language model are you using?" and here is the response I got today: "I am using a large language model (LLM) called PaLM. PaLM is still under development, but it has learned to perform many kinds of tasks, including those I mentioned above. I am excited to see what I can learn to do in the future!"
Do you know if Google have a platform for users to create their own chatbots, like GPTs or like Poe or Zapier does, with customizable knowledge bases, etc?
@@glasco_ I give them a real workout as a research assistant. Bing is way ahead of bard. Unfortunately they all seem to sometimes have blips. I was using perplexity ai lot at one point until it unexpectedly started lying, and today for the first time in a long time, bing is acting the goat.
Gemini Ultra (we do not have access) outperformed GPT4 in tests performed by Google by few %. I have just tested Gemini Pro - extracting data from medical publications and it performed worse than 3.5. I hope that Ultra will be better - we need some other options than GPT4.
😂😂😂😂
Yea people hyped bard before but it’s output was trash last time i tried it so bad I completely deleted it 😂
Well bard was rushed to market to complete with ChatGPT before it was ready. It’s a lot better now
I agree, for me I find Bard very useful for everyday enquiries and coming up for product description and SEO.@@SkillLeapAI
Oh no I gave it another try a couple of days ago. Still said creepy things and lied.
@@SkillLeapAI gave it another shot and it is indeed better
They really did the Happy Gilmore blue duck joke 😅...
"It's an excellent blue duck. Congratulations, you just passed the 1st grade."
I tried bard immediately, but it's still the same old bard that contradicts itself every moment
- Google introduced "Gemini," an AI that surpasses GPT-4 in benchmarks.
- Gemini is labeled the first Universal AI, excelling in text, code, audio, images, and video.
- Three versions of Gemini are unveiled: Ultra for complex tasks, Pro for varied tasks, and Nano for on-device tasks.
- Gemini Pro powers Google Bard, offering advanced reasoning, understanding, coding, and planning functionalities.
- It outperforms GPT-4 in most benchmarks across text, coding, images, videos, and audio, marking a significant breakthrough in large language models and multimodal AI.
They talked up Bard too
While it's still early days for Gemini, it has the potential to revolutionize the field of AI. It can be used in education, healthcare, the possibilities are endless!
I guess we're going to see gpt 4 with video capabilities released to everyone or at least premium subscribers very soon
Based on other things they are saying, Ultra is not going to be free. They were mentioning something like "Advanced Bard" and also they particularly stated that pro was made because that's what's scalable enough to put into general web use. So Pro is supposed to be their GPT3.5 while Ultra would probably be subscription based
Yea I could see them going that direction. But Google has also had the ad base business model. So maybe they will do something similar to that
@@SkillLeapAI I don't think they will. Their statements about having a pro model because that's scalable for general use on the web suggest that Utlra is not scalable for general use on the web and they only have capacity for the few who would pay.
In fact I expect it to have similar bottleneck that GPT-4 is having. Even to the select few who paid for it, GPT 4 is only 40 messages per three hours. What magjc is Ultra gemini using that it's suddenly possible to supply that to everyone who wanted it. I don't think they've made so much progress over the status quo that they can afford to serve something as powerful as GPT-4 to anyone that wants it.
Just prompted Bard with "What language model are you using?" and here is the response I got today:
"I am using a large language model (LLM) called PaLM. PaLM is still under development, but it has learned to perform many kinds of tasks, including those I mentioned above. I am excited to see what I can learn to do in the future!"
Yea they don’t release the update to everyone, all at once
@@SkillLeapAI Funny enough I have access to it now lol.
If it's google then I'm in cause google almost never charges us for anything
How can login
It doesn't outperform GPT-4 since it's the usual Google Undelivered Hype.
I only have the data from their benchmarks. I’ll test it in detail once they roll out ultra.
If we can create AI like this there are limitations
Just tried it,,, disappointment , nice marketing hyped video from G but not there yet
Do you know if Google have a platform for users to create their own chatbots, like GPTs or like Poe or Zapier does, with customizable knowledge bases, etc?
No I don’t think they do yet. Microsoft has something called azure OpenAI service
Chat-gpt is the best
Bing is just gpt online
Bard is the worst and says some creepy things. I think bing has pulled out in front.
Fr? I tried bard and bing 3 months ago and they were terrible and haven’t used either since. Open to retrying bing if it truly has gotten better
@@glasco_ I give them a real workout as a research assistant. Bing is way ahead of bard. Unfortunately they all seem to sometimes have blips. I was using perplexity ai lot at one point until it unexpectedly started lying, and today for the first time in a long time, bing is acting the goat.
More overhyped lol from Google
whats up with that chicks hair/ voice - that google video scared me
its gabriel minus the a, i get it now,,,