Google Gemini, the new era of AI or just hype? 🤖
AI explained simply
Is Google creating a new era of AI? Will this change the future? GPT no more? Has Google overtaken GPT in building the best model? Is this the world's best model? What about Bard? Will Bard be in Google’s graved yard? What is the difference between Bard and Gemini? What’s new in Gemini? Will Gemini beat Chat GPT? When will google launch the Gemini? When Can I try? How much will it cost?….
So many questions right? 🫨 😵 😵💫 🤦♂️.
Hey, this is Madhav Mansuriya again with some new stories on AI, let’s catch up on all of this one by one.
Let’s start with, “Why the name Gemini?”
In April 2023, Alphabet revealed the consolidation of its two artificial intelligence divisions, Google Brain and DeepMind, reflecting the pressure they faced following the success of ChatGPT.
The name Gemini, Latin for “twin,” is directly related to the merger
Statement from Jeff Dean (Chief Scientist, Google DeepMind and Google Research.) Tweet
The Gemini effort came about because we had different teams working on language modelling, and we knew we wanted to start working together —
What is Gemini?
Gemini is a new generative AI model launched by Google on 6th Dec 2023. The tech giant, which calls Gemini the most capable and general-purpose AI it has developed so far, said it plans to expand the advanced version of this large language model (LLM) next year.
Gemini is a multimodal model, meaning it can understand different types of information, including text, audio, images, and video. No need to switch to different models for different kinds of inputs, it’s a one-stop solution.
Gemini will be available in three versions(models): Gemini Nano | Gemini Pro | Gemini Ultra
- Gemini Nano: Gemini Nano is for Android users building Gemini-based apps. It helps summarize recordings from the Pixel 8 Pro’s Recorder app, but only in English (available in Pixel 8 Pro).
- Gemini Pro: Gemini Pro can used for a variety of tasks, this model is currently being served in Bard.
- Gemini Ultra: the largest and most capable, for highly complex tasks
People who have a Pixel 8 Pro can now use a Gemini version for suggested text replies on WhatsApp, and it will also be available on Gboard in the future.
Only a small part of Gemini is out now. In the future, it will do more things, like a chatbot that understands different kinds of input and gives answers in different ways. Right now, only the version that works with text has been added to Bard.
Why Gemini?
Was the Pathway Language Model 2(PaLM 2) not enough for Bard? Following the immense success of ChatGPT’s GPT-4, Google had to take action to stay competitive in the market. LaMDA was launched by Google in 2021.
Google says the upcoming Gemini Ultra performed better than advanced models like GPT-4 in 30 out of 32 benchmark tests, including reasoning and image recognition.
When/Where can I use Gemini?
- The wait is done; a part of Google Gemini is now integrated into Google Bard. Officially, Bard is now utilising a segment of Gemini as its backend model.
Was the Gemini demo Fake? 😲
See what people say:
- The model was not prompted using audio and its responses were only text-based
- The video was crafted “using still image frames from the footage, and prompting via text,”
see what other platforms view about Gemini Demo
Techcrunch, Bloomberg and The Register
Comparison/Difference/Misconception
Is Gemini better than ChatGPT?
- Comparing Gemini and ChatGPT doesn’t make any sense, as Gemini is an AI multimodal model launched by Google and ChatGPT is an app interface that allows you to interact with Open AI’s model GPT 3.5/4
- you can compare the app interface of Bard and ChatGPT, as both are app interfaces that allow you to interact with the AI Model. (Comparing the interface won’t make any difference though 😆)
Will Gemini replace Bard?
- Gemini is an AI model and Bard is an interface to interact with the AI models developed by Google.
- An interface can’t compare AI Models.
- Bard is now powered up with a small chunk of Gemini.
Will Gemini Kill GPT-4?
- Certainly, according to Google’s benchmarks and tests, Gemini outperforms GPT-4 in various tasks. However, its real-world performance is yet to be observed in everyday use.
