What Is Google's AI Project Gemini and What Is It Working On? – MUO – MakeUseOf
Google wants to revolutionize the way we use AI with Gemini.
Google does not cease to amaze. This time, it's here to slay with the Gemini AI project. While Gemini AI is still in the works, there is hope and expectation on the way, considering this project intends to compete with OpenAI's ChatGPT app.
Is Google Gemini a ChatGPT competitor? If the rumors are true, this enormous project will bring forth the best of Google's best minds.
If you are intrigued by Google's Gemini project, here are a few things to familiarize yourself with.
ChatGPT has revolutionized the entire technology industry with its state-of-the-art capabilities and innate ability to play around with text manipulation. Despite ChatGPT's popularity, Google's latest artificial intelligence project Gemini is here to give it a run for its money.
But what is Google's Gemini project? This is what Google DeepMind's CEO, Demis Hassabis told Wired:
At a high level, you can think of Gemini as combining some of the strengths of AlphaGo-type systems with the amazing language capabilities of the large models.
Google's Gemini AI model is a large language model (LLM) that works seamlessly with text. Its prowess resembles GPT-4, the LLM behind the ChatGPT app. In 2016, AlphaGo, an artificial intelligence program, defeated Go's (a board game) world champion.
In April 2023, Google brought together Google Brain, its native deep learning artificial intelligence research unit, and AlphaGo's creators, DeepMind, to create Google DeepMind. Under Google DeepMind's tutelage, the entire conceptualization and development of Google Gemini is taking place.
Google and DeepMind have tried to create a worthy competitor to the ChatGPT chatbot before, with Google Bard competing with ChatGPT. However, with the strong collaboration between the two, Google hopes to create Gemini to beat its competition finally.
The rationale behind Gemini is simple. Google wants to enhance its functionalities for enterprise-based products, such as Google Docs and Slides, and power Google Bard to tackle the competition and make it more resource-friendly.
Additionally, developers must pay Google to access the Gemini AI app via its Google Cloud server-rental division. This move will directly challenge Microsoft and its AI integrations within Office 365 products.
Additionally, with Gemini potentially launching in the fall of 2023, there might even be advanced developments in the field of medical science, as medical AI chatbots and robots could run on Gemini under the hood.
Google has an extensive knowledge base and a rich, data-driven application base, which will be the primary training resource for the Gemini AI application. Add a deep, experience-driven understanding and a flourishing talent pool to this large language model (LLM) training list, and you already have a winner.
Gemini is using a new architecture to merge a multimodal encoder and decoder. The former's job is to convert different data types into a common language, which the latter understands and acts upon. The decoder further generates outputs in various modalities, which is the winning point for the latest Google invention.
You can expect novel outputs with Gemini since it is not dependent on its base data training only. The application will be able to evolve as it progresses without needing to rely on its original data training.
Will ChatGPT and Google's Gemini be at loggerheads once the latter rolls out? Well, Google does intend to snuff out its competition with the launch of its newest application.
At the time of this writing, here's what's known about Gemini and how it fares with its existing competition, ChatGPT.
GPT-4 is a large language model with between 1 trillion and 1.7 trillion parameters. It writes essays, translates languages, and quickly answers questions. But, ChatGPT has limitations to what it can and can't do.
On the other hand, once released, Gemini will be a multimodal intelligence network capable enough to handle various tasks, data-driven requirements, images, audio, videos, 3D models, and even graphs. Since Gemini is a culmination of model networks, it can handle multiple requests simultaneously without limiting itself.
Google Gemini's release date is still unknown at the time of writing. It might be available to the wider public between September and December 2023, per the rumors circulating its release.
Google Gemini and OpenAI will battle it out in the open market post Gemini's release later this year. While OpenAI's ChatGPT has garnered enough attention from its users, Gemini AI should come as a breath of fresh air with its new perspective on AI and enhanced capabilities.
So, is Google Gemini a ChatGPT competitor? Only time will tell, as there is still much dependent on Gemini's release and its final list of capabilities.
Sahil has two years of technical writing experience. His love for technology made his foray into this field, and ever since, he has not looked back.
On the personal side, he loves to travel, explore new places, and go on adventures with his two dogs.
