Amazon Web Services Gets a Bunch of New AI, Machine Learning Features

Amazon.com this week announced a flurry of new machine learning features for its Amazon Web Servicescloud computing business, raising its challenge to Silicon Valley's biggest tech firms for the lead in artificial intelligence.

The new offerings will enable AWS customers to develop and quickly "train" their own artificial intelligence algorithms, build software applications capable of translating language on the fly, analyse video, and scan text for trends or key phrases.

Artificial intelligence (AI) refers to machines carrying out tasks that are normally associated with human intelligence. Machine learning (ML) is a subset of AI in which sophisticated computer algorithms are developed to recognise patterns in large volumes of data to solve problems on their own.

For example, with two of the new AWS features a company could quickly transcribe customer phone calls and then analyse the text for customer sentiment.

Already Apple, Facebook, Amazon and other top tech companies are developing and using AI for their own products, but the new offerings from AWS could make it easier and more affordable for startups and less tech-savvy enterprises to implement AI technology.

The product announcements, made at AWS's annual conference in Las Vegas, cap off a year in which Amazon released 1,300 new AWS features, up from a little more than 1,000 in 2016.

"As always, Amazon is making it easier for companies to get started using new technologies," said Mikhail Naumov, co-founder of DigitalGenius, a London-based customer service startup that uses AI. "Now they are making it easier for companies of all sizes to leverage powerful ML tools in their business."

Despite being the pioneer and dominant player in the cloud computing market, AWS is playing catch-up to chief rivals Microsoft Corp and Alphabet's Google when it comes to new AI offerings, several of which will not be generally available until sometime in 2018.

Microsoft, for example, offered Translator, a direct competitor to the new Amazon Translate, as far back as 2011, Microsoft said. And Google Cloud Platform introduced the Google Natural Language API, a rival to the new Amazon Comprehend, last November. Throughout 2017 both Microsoft and Google announced AI services that rival those unveiled this week by AWS.

"If Amazon can offer products that are just good enough, it can use its leading position," said Chris Nicholson, CEO of Skymind, a San Francisco startup that provides AI solutions for enterprises.

Among Amazon's AI announcements is Amazon SageMaker, which lets companies build and quickly train machine learning algorithms. It also announced Amazon Rekognition Video, which uses AI to detect objects and faces in customers' video content; Amazon Transcribe, which turns audio into text; Amazon Translate, which translates text; and Amazon Comprehend, which analyses text for sentiment and key phrases.

"We expect the big three to continue to play a game of leapfrog over the next several years as the enterprise moves from experimental to industrialisation of AI and machine learning," said Ken Corless, a principal in Deloitte Consulting's cloud engineering practice. "Given their market share, AWS's announcements are significant as they are signalling to the market that they will not cede this space to Microsoft or Google."