Google has released its new artificial intelligence (AI) model set to rival ChatGPT – and it looks pretty damn impressive.
The AI is capable of more than simplifying responding to questions in text form. It’s ingrained with reasoning skills, as well as knowledge and information on worldly topics. Furthermore, the model is designed to be “multimodal”, meaning it can simultaneously process text, audio, images, video, and computer code.
For instance, if you showed it a video of you drawing a car, it would be able to quickly recognize what you’re up to and essentially say “yep, that’s a car.” You could then draw another car and ask the model which one would theoretically move faster based on its design and aerodynamics.
A bunch of other examples of Gemini's intelligence can be seen in the demo videos below.
Gemini was tested on its problem-solving abilities and knowledge in 57 subject areas – including math, physics, history, law, medicine, and ethics – and was found to outperform human experts. Google also said Gemini performed at a higher level than other “state-of-the-art” AI models (such as ChatGPT-4) on 30 out of 32 benchmark tests assessing reasoning and image understanding.
The first version, Gemini 1.0, comes in three different sizes – Ultra, Pro, and Nano – which will be used depending on the needs of the program and user. For instance, Ultra will be the most mighty, but it will be unnecessarily powerful for most day-to-day needs.
Bard will start using a version of Gemini Pro from today and Google plans to integrate other forms of the new AI system across other products like Search, Ads, and Chrome over the coming months. Some of the demonstrations seen in this article won't be accessible to most mortals for the time being, but they will eventually be made available to everyone, in theory.
Google’s announcement was keen to stress that Gemini has been created with responsibility and safety in mind. The recent rush of AI technology has left many people, including industry leaders and scientists, concerned that we might be opening Pandora’s Box by creating a system so powerful it could run amok.
While unveiling Gemini and its omnipotent skills, Google attempted to allay some of those fears.
“At each stage of development, we’re considering potential risks and working to test and mitigate them,” Google said in the blog post.
“Gemini has the most comprehensive safety evaluations of any Google AI model to date, including for bias and toxicity [...] We’ll continue partnering with researchers, governments and civil society groups around the world as we develop Gemini.”