Google announced the general availability (GA) of generative AI services based on Vertex AI, the Machine Learning Platform as a Service (ML PaaS) offering from Google Cloud. With the service becoming GA, enterprises and organizations could integrate the platform’s capabilities with their applications.
With this update, developers can use several new tools and models, such as the world completion model driven by PaLM 2, the Embeddings API for text, and other foundation models in the Model Garden. They can also leverage the tools available within the Generative AI Studio to fine-tune and deploy customized models. Google claims that enterprise-grade data governance, security, and safety features are also built into the Vertex AI platform. This provides confidence to customers in consuming the foundation models, customizing them with their own data, and building generative AI applications.
Customers can use the Model Garden to access and evaluate base models from Google and its partners. There are over 60 models, with pals for adding newer models in the future. Also, the Codey model for code completion, code generation, and chat, announced at the Google I/O conference in May, is now available for public preview.
Vertex AI gives builders a full set of tools to help them tune, launch, and manage models in production. For example, it was the first enterprise-grade MLPaaS to offer Reinforcement Learning with Human Feedback (RLHF) in May. This service leverages human feedback to improve the accuracy of fine-tuned models trained with custom datasets. With Generative AI Studio becoming generally available, customers can use a wider range of tools, such as multiple tuning methods for large models, to build custom generative AI applications much faster.
Google also announced case studies and evidence of customers utilizing its generative AI platform. GA Telesis is using the PaLM model on Vertex AI to build a data extraction system that uses email orders to create quotes for customers automatically. GitLab’s “Explain this Vulnerability” feature uses the Codey model on Vertex AI. This capability gives developers a natural language description of code flaws and suggestions for how to fix them. Canva, the online design tool, helps its users who don’t speak English by using Google Cloud’s generative AI to translate languages. It is also trying ways to use PaLM technology to turn short video clips into longer, more interesting stories. Vertex AI is also being used by companies like Typeface and DataStax to build new tools for generative AI.
In other news, Google has also made Enterprise Search on Generative AI App Builder (Gen App Builder) easier to use. This means that companies can use generative AI and Google’s semantic search technologies to make their own chatbots and search engines. The Gen App Builder has out-of-the-box starter kits for popular use cases of generative AI.
Google assures customers that with Vertex AI and Gen App Builder, their data remains under their full control and will not leave their tenant. The data is safeguarded during transit and while at rest, and Google will not share it or use it for training their models. Google tests its new models carefully to ensure they meet its Responsible AI Principles, and all of its generative AI services include the user security, data management, and access controls that Google Cloud customers have come to expect.
Cloud providers are competing in the field of Generative AI, which allows for the creation of new content using machine learning. This gives customers the option to choose from multiple platforms.
Microsoft is positioning itself as a leader in this area by partnering with OpenAI and making significant investments. With Google announcing the general availability of its own generative AI platform, customers get the choice to choose the best option for their specific business needs.
Read the full article here