From my tests so far, Gemini is better than Chatgpt for writing blog posts. Chatgpt always overcomplicate the text .. I always have to ask it to rewrite, using more straightforward text.. It's annoying.
it gave me the wrong answer. Here is my chat log: [ me: Today I have three cars. Last year I sold two cars. How many cars have I got now? gemini: Here's how to solve the problem: Start with: 3 cars Sold: 2 cars Calculation: 3 - 2 = 1 Answer: You have 1 car now. ] @@Kanknar
Let me share my findings. So uploaded an image into gemini, note you can only upload one, unlike chat gpt you can upload more than one. so i've asked the gemini to read the jpg advert and make a simon sinek start with why book principles advertisement. the model comes back with unfortunately i cannot read pictures. then went to chat gpt, uploaded my paid for PDF and the picture of the advertisement, and it created a one paragraph story telling advert using the image details, pricing and wording and the attached pdf book in one nutshell. Google lost the race, let them leave it to ChatGPT (the proffesionals)
and another note, the Chat GPT just listens better and understand your voice better when you dictate via the mobile app, try using that with the Hey google voice, the voice to speech is so bad that its not even usable, the Gemini app cannot even understand my voice properly but when i talk to chatgpt it just gets it right all the time. no again it just shows, leave the AI race to ChatGPT, the rest has missed the mark
They aren't "autocomplete". They >*train*< by guessing the next thing yes, but it doesn't work like autocomplete internally. In fact, to make a good prediction of the next word, sentence or action you need to actually understand what the prompt means. LLM's do, generally, but confabulate stuff sometimes. Autocomplete is just three words combinations that often are seen together, and doesn't really do any computation on semantics. Far from it. It's like calling humans autocomplete because we also "predict" the next thing with out internal workings. :P Or that airplanes are meteorites because they both "fall"
@@makavelismith Fair enough. When google said their Ultra model was better than chat gpt, they must actually test it b4 releasing it. Else they must have admitted they were no good as chat gpt and released it. Fake promises sucks big time !
hmm, what do you think they are waiting? For 2025? Until GPT 5 is released? I don't see any point in waiting longer if OpenAI has been ahead of them for a year now and they have been scrambling to catch up.
@@CosmicCells As I can recall I saw a comparison between ChatGpt 4 and Google's AI model(maybe they call it Gemini ultra pro or something like that) which shows this AI has outperformed Gpt4 and yet to be released(as per my intuition, in first half of 2024). but also there is a concern and also as you mentioned GPT 5 is also coming. Maybe GPT 5 could dethroned that google AI model. let's see who wins.
@@rahulchauhan1597 Yes you recall correctly. That model that they showed to be slightly better in benchmarks is gemini ultra, so the model in gemini advanced that was shown and reviewed here. So this is their card. The benchmarks comparison was also not fair if you look closely that they compared 5 shot prompting with a different method, im not an expert, but experts said it was an unfair comparison to make geminis benchmark look better htan GPT-4s
From my tests so far, Gemini is better than Chatgpt for writing blog posts. Chatgpt always overcomplicate the text .. I always have to ask it to rewrite, using more straightforward text.. It's annoying.
😂😂 I really laughed when you said ''Gemini is now hallucinating''. LOL
It's not available for work accounts. Why would I need 2 TB of storage if it's not a work account?!
Personal photo storage
Anyone want confirmation of how bad it is, ask it "Today I have three cars. Last year I sold two cars. How many cars have I got now?"
Omg it couldn't even get that right smh. Google has no business charging any money for this. I'll be sticking with ChatGPT 4.
It provides a completely fine answer for me
it gave me the wrong answer.
Here is my chat log:
[
me: Today I have three cars. Last year I sold two cars. How many cars have I got now?
gemini:
Here's how to solve the problem:
Start with: 3 cars
Sold: 2 cars
Calculation: 3 - 2 = 1
Answer: You have 1 car now.
]
@@Kanknar
If you are not getting the right answer just tell it to take a deep breath and to read the question again, believe me it works 😂
ah you really need to learn how to use any LLM
Let me share my findings. So uploaded an image into gemini, note you can only upload one, unlike chat gpt you can upload more than one. so i've asked the gemini to read the jpg advert and make a simon sinek start with why book principles advertisement. the model comes back with unfortunately i cannot read pictures. then went to chat gpt, uploaded my paid for PDF and the picture of the advertisement, and it created a one paragraph story telling advert using the image details, pricing and wording and the attached pdf book in one nutshell. Google lost the race, let them leave it to ChatGPT (the proffesionals)
One thing is for sure: Gemini is fast! 😂
I unsubbed fast.
and another note, the Chat GPT just listens better and understand your voice better when you dictate via the mobile app, try using that with the Hey google voice, the voice to speech is so bad that its not even usable, the Gemini app cannot even understand my voice properly but when i talk to chatgpt it just gets it right all the time. no again it just shows, leave the AI race to ChatGPT, the rest has missed the mark
They aren't "autocomplete". They >*train*< by guessing the next thing yes, but it doesn't work like autocomplete internally. In fact, to make a good prediction of the next word, sentence or action you need to actually understand what the prompt means. LLM's do, generally, but confabulate stuff sometimes. Autocomplete is just three words combinations that often are seen together, and doesn't really do any computation on semantics. Far from it. It's like calling humans autocomplete because we also "predict" the next thing with out internal workings. :P Or that airplanes are meteorites because they both "fall"
So the last prompt to Chatgpt4, did it work?
Can it create AI videos? as it says up to one hour?
Nope.
Well it kind makes sense. Seems like soft opening of Gemini. That is why its forbidden for Workspaces.
It's a Trainwreck. Ive already unsubbed.
you guys keep paying a monthly subscription fee for gd demo lol
Really disappointing from google.
agreed. This product is crap. Do they really think this compares to gpt 4 ? They are way behind. very surprising
Google fked up big time from what i see now. Anyway let's wait and see
@kavinvikram290 nah I'm washing my hands with their false promises.
@@makavelismith Fair enough. When google said their Ultra model was better than chat gpt, they must actually test it b4 releasing it. Else they must have admitted they were no good as chat gpt and released it. Fake promises sucks big time !
I think googl is waiting to show his cards
hmm, what do you think they are waiting? For 2025? Until GPT 5 is released? I don't see any point in waiting longer if OpenAI has been ahead of them for a year now and they have been scrambling to catch up.
it is proven by Google that releasing 1st doesn't make u better proven by Google Chrome@@CosmicCells
@@CosmicCells As I can recall I saw a comparison between ChatGpt 4 and Google's AI model(maybe they call it Gemini ultra pro or something like that) which shows this AI has outperformed Gpt4 and yet to be released(as per my intuition, in first half of 2024).
but also there is a concern and also as you mentioned GPT 5 is also coming. Maybe GPT 5 could dethroned that google AI model.
let's see who wins.
@@rahulchauhan1597 Yes you recall correctly. That model that they showed to be slightly better in benchmarks is gemini ultra, so the model in gemini advanced that was shown and reviewed here. So this is their card. The benchmarks comparison was also not fair if you look closely that they compared 5 shot prompting with a different method, im not an expert, but experts said it was an unfair comparison to make geminis benchmark look better htan GPT-4s