Google Cloud offers several new and enhanced services and features that may help businesses take better advantage of artificial intelligence and machine learning. These new services are already in beta, and these include the AI Platform, improved AutoML solutions, and third-generation Cloud Tensor Processing Units (TPUs).
The AI Platform is a service that allows developers and data scientists to collaborate on artificial intelligence projects more efficiently through a shared interface. The AI Platform features a labeling system that classifies the images, text, or other content supplied by the researcher. After identifying the type of data, the developer of AI models may develop their custom models or they may feed the information into Google’s AutoML, which uses Google's existing technologies to create machine learning (ML) models more quickly and cost-effectively.
The AI Platform also supports Google’s open-source platform Kubeflow. This platform allows businesses or developers to develop the models on servers within the laboratory’s premises rather than on Google Cloud’s servers, which may be necessary for tech firms or government agencies handling sensitive data. Moreover, Kubeflow Pipelines not only permits developers to manage and share experiments or models, but it also helps scientists to better explain the models to individuals who are not as knowledgeable of AI, like businesspeople.
Aside from building the infrastructure that scientists need in developing AI models, Google Cloud also announced improvements to its existing AutoML feature, which allows smaller companies to create models despite the lack of financial and human resources. These three new features are the AutoML Table, the AutoML Vision Edge, and the AutoML Video.
The AutoML Table service enables companies to create models from large volumes of data already arranged in tables without learning complicated or specialized code languages. Meanwhile, the AutoML Vision Edge helps companies create models optimized for devices located at the edge, which may include sensors and cameras.
For businesses, models developed using AutoML Vision Edge can detect defects in products within the factories. The AutoML Video, on the other hand, automatically classifies media content, and media firms may use the models from this service to quickly remove commercials and create short teasers or highlight videos for television programs or film projects.
All these artificial intelligence features and services are supported by Google’s Cloud TPU, which are processors explicitly designed for artificial intelligence tasks. The search giant launched the third-generation liquid-cooled version of its Cloud TPUs back in May 2018, but Google Cloud recently announced that its latest AI processors are now available for use by third-party developers.
Google Cloud is actively developing artificial intelligence services, which help the service differentiate itself from rival offerings from Microsoft and Amazon. For example, AutoML has been available to consumers since early 2018, and Google continues to improve the feature through enhancements made to existing technologies like AutoML Vision and AutoML Natural Language, allowing models to identify position of objects, identify company-specific keywords or phrases, and better understand opinion or emotions expressed in an article or a post.
Developing infrastructure for artificial intelligence helps companies and data scientists that lack the required resources for machine learning projects to create models for their use. Aside from assisting third-party developers, these new features could help Google Cloud snatch some of the customers from its rival services.