Google Gemini

Subject : Science and Tech

Section: AWARENESS IN AI

Context: Google has introduced Gemini, a new multimodal general AI model

What is Google Gemini:

Gemini is a family of highly capable multimodal models, offering features such as image recognition and real-time speech, reported to be five times more powerful than GPT-4.
Gemini is now accessible to users worldwide through platforms such as Bard, various developer platforms, and the newly released Google Pixel 8 Pro phones.
The AI model comes in three sizes: Ultra (yet to be launched), Pro, and Nano.
Gemini is positioned as Google’s response to ChatGPT, which has been a frontrunner in the field of generative artificial intelligence (GenAI)

How Gemini has performed:

Google shared a comparison sheet showcasing how Gemini surpasses GPT-4, the most powerful model of ChatGPT.
Gemini Ultra, designed for data centers, achieved state-of-the-art benchmarking and is under safety review, expected to be available in early 2024.
Gemini excels in Massive Multitask Language Understanding (MMLU) with a 90% score, covering 57 subjects across various domains.
The model is trained on massive datasets, including text, code, books, and multimedia, enabling real-time learning and constant improvement.
Gemini’s capabilities in analyzing vast data, recognizing patterns, and generating hypotheses are seen as revolutionary for scientific research.
Gemini Pro, integrated into Google’s chatbot Bard and other apps, outperforms GPT-3.5 in six out of eight benchmarks, making it a powerful free AI chatbot.
Gemini Ultra, the most powerful in the family, outperforms GPT-4 in 30 out of 32 academic benchmarks, showcasing superiority in reasoning and code generation.

What is ChatGPT:

OpenAI is the artificial intelligence research company responsible for developing ChatGPT, an AI chatbot employing natural language processing to generate humanlike conversational dialogue.
This language model can answer questions and generate various types of written content, encompassing articles, social media posts, essays, code, and emails.
As a form of generative AI, ChatGPT enables users to input prompts and receive AI-created images, text, or videos that resemble human output.
The term “GPT” in ChatGPT stands for “Generative Pre-trained Transformer,” reflecting the model’s process of handling requests and crafting responses.
Reinforcement learning, incorporating human feedback and reward models that rank responses, is used to train ChatGPT, enhancing its machine learning capabilities for improved future interactions.
Besides ChatGPT, notable examples of AI algorithms, also recognized as AI Chatbots, include Google’s Bard and Microsoft’s Bing Chat.