Google Cloud Next Keynote Recap For AI: Translation Hub, Vertex AI Vision & More
Google Cloud Next started today with their keynote presentation. In it, some new AI products were shown off.
Created on October 11|Last edited on October 11
Comment
Google Cloud Next, Google's multi-day globe-spanning conference for Google Cloud products and offerings, has wrapped up the keynote presentation.
In the keynote, they showed off many new features and the ways different companies are using the platform to grow their businesses. A few new AI offerings were also shown off, including new translation tools and vision applications.
You can watch the full keynote here:
Translation Hub
The first AI announcement was Translation Hub, a translation tool designed to translate entire documents into many languages while preserving style and layout. Translation Hub is designed for businesses or organizations that can't or don't want to rely on outside translation services which can become costly and take a long time at large scales.
With Translation Hub, documents can be translated into 135 different languages leveraging the AI that powers Google Translate. The design and structure of documents are preserved so that the document appears as intended across all languages.
Even the best AI translation models can hit hiccups, struggle with certain phrases, or even just get something wrong. To highlight potential mistakes, Translation Hub includes an interface for making corrections which contains a list of each translated sentence along with its confidence score. In there, human translators can quickly identify sentences that might have errors and fix them right away.
Vertex AI Vision
The other main AI announcement today was the reveal of Vertex AI Vision, an expansion of their Vertex AI platform. Vertex AI Vision, like its base application, lets you create and manage AI systems from an easy-to-use interface, but this time with a focus on vision tasks.
Pretrained models will let you create AI-based applications without needing to code, and data output can be streamed into your preferred data management application.
Envisioning an AI-powered collaborative workspace
Towards the end of the keynote, while talking about Google Workspace, they demoed a new vision for how AI could work with teams to generate high-quality content for their brand, created automatically with natural language instructions.
In the demo, the human selects and modifies the AI output with simple controls until it starts to look like the product they envisioned. Once it's tweaked, they send it to the rest of the team for feedback, and single-click suggestions show up thanks to AI.
There's no real product attached to this showcase, so it remains hypothetical, but it's not impossible to think AI could show up in a role like this in the future.
Strengthened partnership with NVIDIA
Google announced a "deeper strategical partnership" with NVIDIA while talking about AI infrastructure on Google Cloud.
Google will be relying even more on NVIDIA GPUs for hardware compute, as well as their other AI services. They are also partnering to promote open-source and open collaborative frameworks such as openxla, JAX, and monai.
Some more AI mentions
While talking about computation infrastructure on Google Cloud, they announced that TPU v4 processing units are now generally available.
Find out more
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.