Artificial Intelligence (AI) is the newest, coolest kid on the IT services block and few companies offer more innovative designs directly to their customers than Google Cloud Platform (GCP). Behind various Google services lies the powerful Cloud AI, a neural net-based, machine learning technology that Google has perfected for use with services like image search and voice recognition. Now, Google makes those capabilities available to businesses. Google Cloud Platform Cloud AI offers robust features and services to meet countless business needs. Here are five ways you could use Cloud AI to improve your business:
1. Improve internal development and innovation with TPU, Cloud AutoML, or ML Engine
TPU, or Tensor Processing Unit (currently in Beta), is used in place of, or in addition to, CPU or GPU processing for machine learning. It was developed specifically for use with TensorFlow, Google’s open-source machine learning platform. TPUs accelerate machine learning and are between 15x and 30x faster than CPUs or GPUs, offering up to 180 teraflops of compute power and 64 gigabytes of memory.1 Since TPUs are cloud-based, you have instant access to needed power for performing mission-critical machine learning training processes. Lyft, for example, uses them to develop self-driving cars, a process that requires vast data sets.2
TPUs may provide power and speed, but Cloud AutoML offers flexibility and accessibility for inexperienced developers. It is a services suite designed to help developers train high-quality machine learning models using Google’s existing APIs. AutoML Vision is currently available, but more products are on their way in every significant AI field. AutoML Vision allows customers to train machines to categorize products better, making customer searches more accessible.
For experienced ML developers, Google offers Machine Learning (ML) Engine. ML Engine is a managed service for bringing machine learning models to production. It provides both training and prediction services. You can supply TensorFlow models to be trained by the service for several scenarios. The prediction service takes trained models and uses them to make predictions on new data. This service can generate either online (real-time) predictions or less-expensive batch predictions.
2. Improve image and video processing and identification with Cloud Vision API and Cloud Video Intelligence API
The Cloud Vision API provides identification features that businesses could utilize in various ways. The API can identify objects, logos, and landmarks within images for more natural categorization or identification. It can also detect explicit content, making it easier for web-based companies to identify and remove objectionable content, including text within an image, using Optical Character Recognition (OCR). It can find similar images on the web as well, or detect faces and read expressions to provide insights regarding customer mood. It does not, however, provide facial recognition.
Similarly, Cloud Video Intelligence API allows you to search and find videos by extracting metadata and identifying and cataloging video content, without any machine learning expertise. It also allows users to decide the most relevant content by providing confidence levels for each item identified. It can locate objects in an entire video, a particular shot, or even frame by frame. The API could be useful to businesses by identifying inappropriate content uploaded to websites in order to block or remove it. Additionally, media archives of all kinds, from academic to commercial, could use this API to speed metadata processing, making it instantly available while saving employee time.
3. Improve communications both with customers and among employees using DialogFlow and Natural Language API
DialogFlow can handle customer messaging, voice recognition and response, using applications such as chat bots, for help and search features online. DialogFlow is a development suite that allows you to build interfaces for mobile apps, messaging and IoT devices. It can understand interactions between customers and your business and provide responses. You could also use it to facilitate internal communications. For instance, a retail sales company could use it both for customer service and commerce transactions, while large enterprises may find it more useful for improving employee communications using chat features or for creating smart interfaces for IoT devices. DialogFlow includes a code editor, but you can also use your own webhook.
The Natural Language API can provide even more insight about customer interactions. It can identify syntax, entities, sentiment, and classify content in numerous text languages. You can also combine the API with other services to gather voice data or data from scanned documents. You access it through a REST API and can upload content directly or use cloud storage. Data scientists could use it to expedite the time it takes to analyze large data sets from sources such as customer service chats and social media feeds, to understand better how consumers respond to your products or services.
4. Provide foreign language support for global business with Cloud Speech-to-Text, Text-to-Speech and Cloud Translation APIs
Google's language processing and translations services offer solutions for a host of language compatibility issues. For example, they could be used in call centers, for video captioning, or in services catering to the hearing or vision impaired. The Speech-to-Text API enables speech conversion, either in real time or from recordings. It recognizes 120 languages. The Text-to-Speech API produces natural-sounding audio from text. It offers more than 30 voices and several languages. You probably already interact with IoT devices such as Google Home or similar tools that accept and emit voice commands and responses in this way.
Cloud Translation API can be combined with either of the above, work with other APIs, or stand alone. It provides translation services for over 100 languages. It can deliver language detection if the users do not know the languages they are translating. It gives HTML feedback, but can be combined with Text-to-Speech API if audio is needed. This service could be invaluable to global businesses that require translation services.
5. Make employee hiring processes easier with Cloud Job Discovery
Job Discovery seeks to understand better how job seekers, as well as employers, use and interact with language on job sites. It can provide insights enabling employers to continuously adapt job listings and strategies to find the most qualified candidates.
Eve Eiler contributed to this post.
Never miss another article. Sign up for our newsletter.
1. Stephanie Condon, "TPU is 15x to 30x faster than GPUS and CPUs, Google says," ZDNet, April 5, 2017, https://www.zdnet.com/article/tpu-is-15x-to-30x-faster-than-gpus-and-cpus-google-says/#ftag=CAD-00-10aag7e; Liam Tung, “GPU killer: Google reveals just how powerful its TPU^2 chip really is,” ZDNet, December 14, 2017, https://www.zdnet.com/google-amp/article/gpu-killer-google-reveals-just-how-powerful-its-tpu2-chip-really-is/.
2. John Barrus and Zak Stone, “Cloud TPU machine accelerators now available in beta,” Google Cloud Platform Blog, https://cloudplatform.googleblog.com/2018/02/Cloud-TPU-machine-learning-accelerators-now-available-in-beta.html.