8 Biggest AI Features and New Capabilities for Google Cloud

Explore Google Cloud's 8 powerful AI features, including Google Bard with New Capabilities

Google is intensifying its focus on artificial intelligence and generative AI, unveiling a range of cutting-edge capabilities and software for flagship offerings like Google Workspace and Google Cloud. At Google I/O 2023, CEO Sundar Pichai shared the company's aspiration to lead the AI and generative AI market. This event saw the introduction of innovative products and features, marking Google's dedicated journey towards AI excellence.

Google, valued at $280 billion, is heavily investing in infusing AI across its extensive portfolio of cloud computing, collaboration, and consumer services. Sundar Pichai emphasizes AI's significance and envisions a collaborative effort to harness its potential. Highlights from Google I/O 2023 include Duet AI, Vertex AI enhancements, and Google Bard improvements, as covered by CRN.

These include:

1. Duet AI For Google Workspace
2. PaLM 2
3. New Bard Capabilities
4. Duet AI For Google Cloud
5. Vertex AI New Foundation Models
6. Vertex AI RLHF
7. Google A3 Supercomputers With Nvidia H100 GPUs
8. Vertex AI Embedded APIs

1. Duet AI For Google Workspace

Duet AI for Google Workspace headlines Google I/O, merging robust generative AI functionalities for enhanced collaboration. Gmail mobile integrates generative AI to aid content creation.

Google Slides incorporates AI for the effortless creation of visuals from minimal input. Sheets employ generative AI for data analysis, automated classification, and custom table creation.

 Smart canvas capabilities, like smart chips, elevate assisted writing in Google Docs.

PaLM 2: Google's Next-Generation Language Model

Google introduced PaLM 2, an enhanced version of its large language model.

PaLM 2 encompasses models with distinct sizes: Gecko, Otter, Bison, and Unicorn. Gecko, designed for mobile devices, boasts interactive offline capabilities. CEO Pichai highlighted their prowess in logic, reasoning, and multilingual understanding, having been trained in scientific, mathematical, and over 100 language texts.

3. New Bard Capabilities

At Google I/O, updates to Bard were unveiled, leveraging Google's new PaLM 2 large language model. Bard's evolution includes enhanced visual responses and prompts. Users can now inquire about travel recommendations and receive not only text but also rich visuals for better insights.

 Users can also infuse images into their prompts, boosting creativity. This integration is enabled by harnessing the capabilities of Google Lens, marking an innovative stride in Bard's capabilities.

4. Duet AI for Google Cloud

Google is rolling out Duet AI for Google Cloud, an AI-powered collaborator catering to diverse skill levels. This expert pair programmer aids cloud users by providing contextual code completion, tailored suggestions, real-time function generation, and code review assistance.

Duet AI fundamentally reshapes how users craft experiences across Google Cloud, seamlessly integrated into interfaces like IDEs, Google Cloud Console, and chat, promising enhanced efficiency and innovation.

5. Vertex AI New Foundation Models

Google introduced three new foundation models in Vertex AI, accessible via API, UI in Generative AI Studio, or data science notebooks.

These new foundation models are: Codey, Imagen, and Chirp.

Google's Codey is a text-to-code foundation model, embedded in SDKs or apps to boost developer efficiency with real-time code completion and generation.

Imagen, the text-to-image model, generates and customizes studio-grade images at scale for diverse business needs.

 Chirp, their speech-to-text foundation model, fosters inclusive engagement with customers through native language captioning and voice assistance, enhancing interactions.

6. Vertex AI New RLHF

Introducing New Reinforcement Learning from Human Feedback (RLHF) to Vertex AI, enhancing tuning and prompt design by integrating human feedback for tailored model advancement.

 Google pioneers as the first cloud hyperscale provider to offer RLHF as a managed service within Vertex AI, promoting cost-efficient, consistent model performance and safer, more valuable model deployment to production.

7. Google A3 Supercomputers With Nvidia H100 GPUs

Google Cloud broadens its compute offerings through the private preview launch of the A3 GPU supercomputer, merging Nvidia H100 Tensor Core GPUs with advanced networking.

This expansion caters to various-sized customers, extending compute options for machine learning tasks like large language models, generative AI, and diffusion models.

8. Vertex AI Embedded APIs

Google integrates text and image APIs into Vertex AI, empowering developers to construct advanced applications like recommendation engines and classifiers. Embeddings transform text and image data into numerical vectors, preserving semantic relationships.

 These vectors are efficiently processed by large models and prove beneficial for extensive inputs, such as lengthy texts with thousands of tokens.