Tuesday, May 14, 2024
HomeSci/TechHere is 'GEMINI'- Google's Next Generation AI 1.0 is Amazing

Here is ‘GEMINI’- Google’s Next Generation AI 1.0 is Amazing

Google unveiled Gemini, a next-generation AI model, on December 6. 

Gemini is a next-generation AI model that will power the company’s AI features and services, including the Duet AI assistant and the Bard chatbot. According to the company, Gemini is far more sophisticated and capable of handling more complex scenarios than both Google’s earlier generation AI models and models created by rivals like OpenAI. To bolster these assertions, Google released a number of videos showing off the applications of Gemini.

Google displays a user engaging with Gemini across a variety of input methods and languages. The user can be seen conversing with Gemini, demonstrating hand gestures, pictures, drawings, and asking the model different kinds of questions.

Additionally, you can watch the model handle challenging tasks like multimodal conversation, logical and spatial reasoning, making connections, solving visual puzzles, translating images, and demonstrating cultural understanding.

Researchers from Google DeepMind explain how a dataset of academic papers on genetics was updated using Gemini.

Google unveiled Gemini
Google unveiled Gemini

Tens of thousands of papers were reviewed in order to create the original dataset, and the pertinent information was manually extracted. But since 2021, more than 200,000 new papers have been added to this domain, making it impossible to manually update the dataset. Gemini was able to read through the new papers and extract the relevant information. It was also seen that it can update any graph from the study.

Gemini can help kids improve Academics

A Google engineer demonstrates how a handwritten physics homework sheet can be checked for errors and how customized explanations for each question can be generated using Gemini’s multimodal capabilities and sophisticated reasoning.

Additionally, it can offer practice questions to help assess knowledge, broaden it, or strengthen any areas where it is lacking.

Gemini can Code advanced

Researchers from Google DeepMind show off the model’s sophisticated coding skills by swiftly creating a web application to explore London’s train stations.Additionally, the model is shown to be able to solve theoretical computer science and complex math competitive programming problems. For instance, Gemini once solved a competitive coding problem that was beyond the capabilities of less than 0.2% of contestants. Go, Java, C++, and Python are all compatible with Gemini.

Versions of Gemini 1.0- The very first version of this AI model

The first version comes in 3 sizes-

Gemini Nano-The lightest model, Nano, is made to operate on-device applications on portable electronics, such as mobile phones. It will be capable of operating without internet access. Initially, Gemini Nano will power the Gboard keyboard’s Smart Reply feature and the Recorder app’s auto-summarization feature on the Pixel 8 Pro.With Android 14, starting with Pixel 8 Pro devices, Google said that “Android developers will also be able to build with Gemini Nano, our most efficient model for on-device tasks, via AICore, a new system capability.”

Gemini Pro-The majority of free, user-facing AI services will be built on this model, which is geared toward most general AI tasks. Although Google Bard has been updated to operate on Gemini Pro, an improved version known as Bard Advanced, which runs on Gemini Ultra, is anticipated to debut in the upcoming year.

Gemini vs ChatGPT and Others

Gemini Ultra-For extremely complex tasks, this is the largest and most capable model available.This model will probably support Google’s paid AI offerings or be made available to enterprise customers on the Google Cloud Vertex platform because it requires a lot of resources to run. Before making Gemini Ultra widely available, Google stated, “We’re currently finishing up extensive trust and safety checks, including red-teaming by trusted external parties, and further refining the model using fine-tuning and reinforcement learning from human feedback (RLHF).”

According to Google, Gemini outperforms other AI models in use today, such as OpenAI’s potent GPT-4 model, which powers ChatGPT Plus and Enterprise.

RELATED ARTICLES

2 COMMENTS

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments