Google on Tuesday announced several updates to its Vertex AI platform and an updated version of its text-to-image model Imagen 2.
At the Google Next '24 conference in Las Vegas, the cloud provider revealed that its new LLM, Gemini 1.5 Pro, is available in public preview on Google's enterprise AI platform Vertex AI.
A new version of the image generation model, Imagen 2, now lets you create 4-second live images from text prompts and adds new image editing features.
Vertex AI also added new grounding features, including the ability to ground responses in Google Search. We also added new instant management and evaluation services for larger models. Grounding is an additional step to ensure that the data is accurate and that reliable responses and models are based on something other than the data on which they were trained.
With these new updates and advancements, Google continues the pattern it set at the beginning of the year of advancing its GenAI technology despite increased competition.
According to Forrester Research analyst Rowan Curran, the latest update to Vertex, centered around Gemini 1.5 Pro, announced in February, will allow Google to provide tools for businesses to build around Gemini. He says he is aiming for it.
“Being able to have a model with such a huge context window changes the types of use cases and applications that can get away with it,” Curran said, referring to Gemini Pro's 1 million large context windows option. Did.
Without tools to quickly manage and test new responses, it's difficult for companies to build around models like Gemini, he added.
“This is a whole body of tools that are being developed and deployed specifically around supporting generative AI,” he said.
A variety of tools and capabilities support the Gemini family of models, opening new possibilities for how companies apply and build generative AI, he continued.
Vertex AI updates
One way Google is supporting this is by building new prompt management and assessment services into Vertex AI for large models such as Gemini 1.5 Pro.
The new service will allow users to organize, track, and modify prompts for machine learning models.
“This benefit streamlines the process of creating, editing, and managing prompts,” said Paul Nashawaty, an analyst at Futurum Group.
In addition, rapid management and evaluation services are important for companies looking to build GenAI applications because they can evaluate previous prompts and responses, Curran said.
“We need the ability to log these queries and responses so we don't have to regenerate them exactly in the future,” Curran added.
Meanwhile, Vertex AI's new grounding capabilities, currently in preview, allow users to ground LLM responses in Google Search or enterprise data sources using search enhancement generation. RAG optimizes the output of LLM.
The new grounding feature promises to reduce LLM hallucinations, Nashawaty said.
“That's the main progress,” he said. “That way, companies can confidently increase the use of his LLM.”
Gartner analyst Sid Nag says grounding allows companies to ensure that what their AI systems understand or interact with in the real world is accurate.
“This is a bridge between abstract AI concepts and practical concrete outcomes,” he said.
He added that this not only provides real-world accuracy, but also adds human sentiment analysis to help users avoid errors that can be caused by simulated data.
Curan said the grounding technology comes as more companies look to leverage Google search to support grounding of generative models.
“More and more companies are looking to root the responses of large-scale language models into their data,” he said.
RAG's popularity has led to recent developments from Google's competitors, including Microsoft, which recently announced changes to Azure AI Search that allow customers to run RAG at any scale.
Companies looking to build future AI applications will need predictive AI to understand how likely a customer is to act, generative AI to understand customer intent based on natural language or images, and the right information. You'll need a search tool to retrieve or find it, he said. Said.
“I see the future of this as a great trifecta or tripod built around prediction, generation, and search,” he continued.
Other updates to Vertex AI include pro features in Gemini 1.5 to process audio streams, including the audio portion of voice and video.
Google also revealed that the Anthropic Claude 3 family of models is now available on Vertex AI.
Open models such as Llama 2, Mistral 7B, and Mixtral 8 are also available in Vertex AI.
New features in Imagen 2
Introduced in preview, Imagen 2 includes a text-to-live image feature that allows marketing teams to generate GIFs and video loops from text prompts. Imagen 2 also has advanced photo editing features.
Imagen 2 comes after Google suspended new image generation features in its Gemini conversational app (formerly known as Bard) after the app generated inaccurate images of historical figures.
This is a whole body of tools being developed and deployed specifically around supporting generative AI.
Rowan CurranForrester Research Analyst
Benefits of Imagen 2's live image capabilities include speeding up the creation process, reducing human error, and the ability to minimize tedium, Nashawaty said.
Disadvantages include a steep learning curve for users and high implementation costs, he said. And there are similar models on the market.
“This is not any technological advancement,” Curran said.
For example, Nag points out that open source tools such as the Text2Live tool and Stability AI's Stable Diffusion 3 have similar capabilities.
He added that text-to-live images can also come with privacy issues.
“From a privacy perspective, I don't see any limitations or usage of the Text-to-Live feature,” Nag said. “If that functionality is limited to certain types of enterprise-grade workloads, that's a good thing.”
Google also revealed new partner infrastructure and partner news.
Cloud TPU v5p, the vendor's next-generation accelerator for training GenAI models, is now generally available. Powered by Nvidia H100 Tensor Core GPU.
Nvidia Blackwell GPUs are also coming to Google Cloud.
Nvidia and Google are working together to help startups create GenAI applications and services. Nvidia Inception members can now use Google infrastructure.
All the latest information from Google shows that the GenAI race continues to heat up.
For companies, it's important to focus on GenAI applications rather than introducing the latest products, Curran said.
“It's really important to understand how to build things and what the emerging best practices are,” he said.
Esther Ajao is a TechTarget editorial news writer and host of a podcast covering artificial intelligence software and systems.
This website uses cookies to ensure that you get the best experience on our website. OkRead More
Privacy & Cookies Policy
Privacy Overview
This website uses cookies to improve your experience while you navigate through the website. Out of these cookies, the cookies that are categorized as necessary are stored on your browser as they are as essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may have an effect on your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
