- \n
- Gemini — an update for Google’s Bard chatbot that expands the neural network’s capabilities.
- The tech giant’s AI development outperforms GPT-4 in performance and accuracy.
\n
\n
\n
\n\n
Google has released a global update for the Bard chatbot designed to significantly expand the capabilities of the neural network. The update is named Gemini.
\n\n
We believe in making AI helpful for everyone. That’s why we’re launching Gemini, our most capable model that’s inspired by the way people understand and interact with the world. #GeminiAI pic.twitter.com/gNG9ha9xMO
— Google (@Google) December 6, 2023
What Gemini Can Do
\n\n
Google DeepMind CEO Demis Hassabis and Alphabet CEO Sundar Pichai described Gemini as \u201ca significant leap forward in chatbot technology.\u201d They highlighted its \u201csuperior capabilities\u201d for providing accurate responses and its ability to reason at a human level in certain situations.
\n\n
The updated neural network gained expanded capabilities for helping with code-writing and mathematical calculations, and has learned to recognise images, video and audio.
\n\n
Gemini is available in three variants: Ultra, Pro and Nano. This variety allows the neural network to be tailored to any device for specific tasks—from mobile phones to data centers.
\n\n
Deployment will occur in two phases. Bard will initially update to the free version Gemini Pro. Next year Google will introduce Bard Advanced, which will give users access to the enhanced AI model Gemini Ultra priced at $20 a month.
\n\n
Gemini Nano — a special mobile version that can operate offline. Pixel 9 Pro owners can already test the chatbot.
\n\n\n\n
The early iteration of Gemini Pro will rely on text prompts, but in the coming months its functionality will be expanded to multimodal support, including text, images and other media files.
\n\n
With the debut of Bard Advanced in 2024, which will become \u201ca new experience built on the most powerful Gemini model,\u201d the neural network will gain reasoning capabilities and the ability to interact with various types of information.
\n\n
Following publication of the video showcasing Gemini’s capabilities, some Google developers and industry experts said the technology did not perform as well as expected. One company employee told Bloomberg of the \”overstating\” of the AI’s functionality and performance.
\n\n
Oriol Vinyals, the CEO of Google DeepMind, explained that user prompts and outputs in the video are real, and the generation times were \”trimmed for brevity\”.
\n\n
Really happy to see the interest around our \u201cHands-on with Gemini\u201d video. In our developer blog yesterday, we broke down how Gemini was used to create it. https://t.co/50gjMkaVc0
We gave Gemini sequences of different modalities — image and text in this case — and had it respond… pic.twitter.com/Beba5M5dHP
— Oriol Vinyals (@OriolVinyalsML) December 7, 2023
Against the backdrop of the scandal that spread through the media and social networks, Alphabet’s shares fell from $139 to $135 over a few days, according to MarketWatch.
\n\n
\n\n
Testing Gemini’s capabilities
\n\n
Google has long sought to build a product that competes with ChatGPT. For this reason, Gemini has been tested against OpenAI’s language models.
\n\n
According to the developers, even the standard version of the neural network surpasses GPT-3.5, released in November 2022. In six of eight tests, the Pro version outperformed its rival, making it the most capable free AI-powered chatbot.
\n\n
GPT-4 has already managed to hold its own, but Gemini still outperformed it on several metrics.
\n\n
\n\n
Gemini Ultra showed slightly better results than GPT-4 in object recognition tests and in answering questions. Google’s technology also leads in mathematical reasoning and Python code generation.
\n\n
Additionally, Gemini passed 30 of 32 State-of-the-Art tests, which assess AI’s ability to learn deeply, process natural language and solve general tasks.\u00a0
\n\n
Demis Hassabis told about the new AI model in June. Even then he stated its superiority over ChatGPT.
\n\n
Earlier, the company introduced Bard’s chatbot. A month and a half later, the firm opened access to the tool to a limited number of users.
\n\n
In September, the company announced new AI-powered features that will speed up YouTube content creation.
\n\n
Earlier, co-founder Mustafa Suleyman stated that the next stage of AI development is interactive bots that can perform tasks for humans.
\n