What is Amazon AI?
Amazon AI is a set of artificial intelligence (AI) services that offer machine learning (ML) and deep learning technologies to Amazon Web Services (AWS) customers. These services are intended to help software developers create AI-based software.
AWS offers a range of pre-trained AI services that help organizations prepare data, as well as build, further train and deploy their own AI and ML models. Amazon AI services integrate with an organization's applications for use cases such as personalized recommendations, automated data extraction, text transcription, and customer experience and business metrics improvement.
How Amazon uses AI
Amazon approaches AI internally using a flywheel strategy, where AI and ML are continually innovated, and those innovations are spread throughout other areas of the company. This strategy helps AWS further build upon, expand and improve its services.
Amazon uses machine learning in several ways, including the development of chatbots, voice recognition, fraud detection and product recommendations. AI and ML are used in Amazon products, such as Alexa's and Amazon's recommendation engine, as well as other business areas, such as in Amazon warehouses.
The Amazon AI suite of services provided for other organizations are based on the same tools it uses. The speech recognition, text-to-speech, image recognition and ML services are scalable and fully managed, enabling a developer to include the technology in an application without having to learn algorithms or manage the supporting infrastructure.
Amazon AI service options
Amazon AI services span across categories, including computer vision, automated data extraction and analysis, language AI, customer experience, business metrics, code and DevOps, industrial AI and healthcare. In more depth, they include the following tools.
- Amazon Rekognition uses facial, object and scene recognition to analyze images and determine their content. It provides users with content moderation, face detection and analysis, face comparison and search, custom labels and video segment detection.
- Amazon Lookout for Vision detects process defects in real time using ML models. This tool helps users spot part damage, identify potentially missing components and find other potential process issues.
- AWS Panorama is designed to improve operations concerning computer vision by integrating local area networks with ML devices and a software development kit (SDK). This helps organizations improve supply chain logistics and optimize traffic management.
Automated data extraction and analysis
- Amazon Textract uses AI to automatically extract data, tables, printed text, forms and handwriting data from a document. It can be used in financial services or healthcare, for example.
- Amazon Comprehend is used to understand possible insights within text-based documents. This is useful for searching product reviews, deriving meaning from legal briefs or processing financial documents.
- Amazon Augmented AI is used to implement a review of ML predictions based on defined requirements, such as including multiple reviewers. This is useful for quickly extracting important data from documents, processing financial service data and integrating human reviews for ML workflows.
- Amazon Lex enables developers to build chatbots using conversational AI in applications. It uses automatic speech recognition to convert speech to text and natural language processing (NLP) to understand spoken instruction. Amazon Lex helps users build voice assistants or virtual agents and bots and automate responses.
- Amazon Polly lets developers add natural-sounding speech capabilities into applications. A developer sends text to Amazon Polly through an SDK or AWS Management Console, and Polly synthesizes it into humanlike speech. Users can generate speech in numerous voices and languages and adjust speaking styles, rate of speech, pitch and volume to interact with customers using a natural voice.
- Amazon Transcribe converts speech to text. For example, it can convert speech from calls and video files to text.
- Amazon Kendra uses ML and NLP to adjust search results across several structured and unstructured content repositories. Kendra can be used to improve internal search and to integrate searches to software-as-a-service applications.
- Amazon Personalize can evaluate customer experience using ML and eases the process of integrating personalized recommendations into applications.
- Amazon Translate converts large amounts of text written in one language to another language. Amazon Translate enables users to translate user-generated content, convert conversations between languages and perform cross-lingual user communication.
- Amazon Forecast predicts business outcomes, such as retail and inventory forecasting, workforce planning and product demand, to help increase customer satisfaction.
- Amazon Fraud Detector helps organizations detect fraud by constructing, deploying and managing online fraud detection models using customized business rules.
- Amazon Lookout for Metrics uses ML to help organizations monitor business metrics and detect anomalies by creating custom alerts to identify their root causes.
Code and DevOps
- Amazon DevOps Guru detects unusual application behavior using ML models with the goal of improving application availability. DevOps Guru can detect abnormal behavior in application metrics, logs and events. This tool is useful for improving availability and the performance of serverless applications, identifying resource limits and limiting recovery time for Amazon Relational Database Service databases.
- Amazon CodeGuru Reviewer automates code reviews using ML. It's also used to detect security vulnerabilities and optimize application performance. Developers can use this tool to find code errors before production, fix potential security issues and continuously monitor their code.
- Amazon CodeGuru Profiler optimizes performance and troubleshoots performance-based issues for applications running on premises, Amazon Elastic Cloud Compute, Amazon Elastic Kubernetes Service, AWS Fargate or AWS Lambda.
- Amazon Lookout for Equipment automatically detects potential issues with equipment behavior. It uses data from local sensors and industrial equipment to tailor an ML model to an organization's equipment. This helps organizations detect equipment anomalies, enabling them to quickly diagnose potential issues.
- Amazon Monitron reduces unplanned equipment downtime using ML and predictive maintenance. It's useful for preventing downtime and saving on potentially costly equipment repairs.
- Amazon HealthLake can store, query, transform and analyze health-based data. It can also pair with Amazon Comprehend Medical for further querying. Organizations can use HealthLake to help improve quality of care and efficiency.
- Amazon Comprehend Medical extracts data from unstructured medical text. This is useful for collecting data from texts such as clinical trial reports and doctors' notes.
Amazon AI and machine learning
Amazon uses both AI and ML in its offered services. For example, Amazon SageMaker is a fully managed cloud ML platform that enables developers and data scientists to build, train and deploy machine learning models for predictive analytics applications.
How is Amazon investing in AI?
Amazon is investing in AI by hiring development managers and software engineers to build AI tools. Amazon also pairs with other companies to further the development of AI. For example, along with Microsoft and Alphabet, Amazon created a partnership with C3 AI -- an enterprise AI software provider that focuses on creating enterprise-scale AI applications. Amazon has also paired with other AI-based companies, such as Hugging Face. Hugging Face is an online service where developers of AI share AI code and models in an open source format. This partnership was formed with the goal of enabling AWS developers to take code from Hugging Face and use it on AWS cloud services.
Learn more about the Amazon and Hugging Face partnership, as well as the growing competition in the AI space.