Google announced it is rebranding its AI chatbot, Bard, to Google Gemini, and the change is now live. The new launch also brings a new Gemini app for Android phones and access to the tech giant’s most capable AI, Google Gemini Advanced.
As Google reports, Gemini is a new AI family built from the ground up to be multimodal, meaning it can understand and process information from different sources, including text, code, audio, images, and video. This is different from Bard, which was a single large language model (LLM).
Visiting the new Gemini website shows a banner detailing the change in branding, but it’s otherwise used in the same way as before. Now, the Bard interface offers access to the diverse capabilities of the Gemini family, with “Bard” becoming the name for the user interface you interact with and “Gemini” referring to the underlying technology powering it.
Now in its first version, Gemini 1.0, Google notes that it comes in three different models:
- Gemini Ultra: The most powerful and largest model, designed for tackling highly complex tasks.
- Gemini Pro: The best all-rounder, able to handle a wide range of tasks while balancing performance and efficiency.
- Gemini Nano: The most lightweight and efficient model, optimized for running on devices with limited resources.
Gaining access to Gemini Ultra means opting to subscribe to Gemini Advanced, which is “designed for highly complex tasks” such as problem-solving, coding, generating new images, role-playing as characters a user describes, and more.
This model is a paid plan, part of the new Google One AI Premium plan. It’s now available, costing $19.99/month in the US, £18.99/month in the UK, and €21.99/month in Europe. Prices in New Zealand have yet to be announced as of writing, but given Google’s relationship with NZ, it could be a while until we see better availability.
So, what advancements can you expect with Google Gemini? Asking Gemini itself, one benefit is that it can reason across multiple languages. For example, given an image and a question in one language, Gemini can answer the question in another language, considering the visual information.
Moreover, it can also write poems, code, scripts, musical pieces, emails, letters, and more in response to a prompt or question. You can also expect more informative responses, even if the questions asked are “open-ended, challenging, or strange.”
As Gemini notes, it is still under development, but it has already been integrated into some Google products, such as Duet AI for Developers and dev tools for Chrome and Firebase. Soon, Gemini Advanced will also support apps like Gmail, Google Docs, and more.
Google Gemini states it has some potential risks
While asking Gemini about itself, it stated that there are, of course, potential risks associated with Gemini. This included the following:
- Bias: Like any AI model, Gemini is susceptible to bias, which could lead to unfair or discriminatory outcomes.
- Misinformation: Gemini could be used to generate fake news or other forms of misinformation.
- Job displacement: As Gemini becomes more capable, it could automate some jobs currently done by humans.
These risks aren’t unheard of, especially since the development of other AI platforms, such as OpenAI’s GPT-4. How Gemini may disrupt jobs done by actual people and how it will spread misinformation is yet to be seen, but its clear scope is larger than Google’s previous AI ventures.
In the meantime, you can chat with Google Gemini by heading to gemini.google.com.