What you need to know
- Sources have allegedly stated Google is giving outside developers a chance to play around with its new Gemini AI system.
- Developers have gained access to a smaller version of Gemini while Google continues to “train” the full version.
- Gemini reaching a limited test could mean the company is getting closer to releasing the new AI model for consumers and for developers for software integration.
Google’s next large AI model is reportedly moving into its next phase of development as the company gathers some early feedback.
According to The Information, Google has started granting access to its Gemini AI system to outside developers (via Reuters). Opening the doors to Gemini in a limited way could mean Google is preparing to work Gemini into its “consumer services and selling it to businesses through the company’s cloud services,” the publication added, citing multiple sources close to the matter.
The scale of Gemini that developers have gained access to is apparently not quite as large as OpenAI’s GPT-4 model — which it is designed to compete with. The report states Google is still in the process of “training” the full breadth of its Gemini AI before giving any outsiders a chance to experiment with it.
Outside developers gaining such an early preview could mean Google will let them grab it through Vertex AI and integrate the new model with their services. Something like this was leaked a couple of months back as it was discovered Google is planning on rolling out Bard Extensions for its AI chatbot. The discovery unearthed the bot potentially seeing integration with Flights, Maps, YouTube, and Zillow, among others.
Gemini is a work in progress; however, Google has stated it is designed to be a next-generation foundation model. The company will look to mirror PaLM 2 in the sense that Gemini will be available in various sizes with varying capabilities depending on where it’s to be used.
Furthermore, Google’s co-founder Sergey Brin returned to the company to help shape its Gemini AI model. While the brand’s current CEO, Sundar Pichai, expressed his excitement, Google’s goal with Gemini is to create a more “human-like” artificial intelligence system.
Google is playing a bit of catching up as it prepares its next major AI system in Gemini. OpenAI has long since rolled out its latest language model, GPT-4, to its users through its ChatGPT monthly subscription service. The model is also available for use on its Android ChatGPT app — iOS, too.
Read more here: Source link