agsandrew - Fotolia
Microsoft and Amazon have each given a boost to developers looking to build applications on their respective artificial intelligence (AI) platforms.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
As businesses start to look seriously into making use of AI, platforms such as Microsoft’s Cognitive Services and Alexa from Amazon will become increasingly used to develop new applications.
Both companies are looking at how to make artificial intelligence more widely available to developers.
Microsoft said its ambition was to democratise AI and make it accessible for all.
This is the term CEO Satya Nadella used in January at the World Economic Forum. “Both the state I was born in in India and the state I now live in in the United States are using statistical machine learning to improve high school outcomes and use scarce state resources smartly,” said Nadella during a panel discussion. “That, to me, is democratising AI.”
AI tools for developers
Microsoft has already made the AI technology behind its Skype Translator, Bing and Cortana available to third-party applications through Microsoft Cognitive Services – a collection of 25 tools which allow developers to add features like emotion detection, vision and speech recognition and language understanding to their applications.
Microsoft has now extended this with the three tools Custom Speech Service, Content Moderator and the Bing Speech API.
Custom Speech Service offers developers speech recognition technology which Microsoft claimed can work accurately in noisy environments, and cope with jargon, dialects and accents.
Content Moderator provides a way for users to quarantine and review data, such as images, text or videos, and filter out unwanted material, such as potentially offensive language or pictures.
The third tool, Bing Speech API, can be used to convert spoken audio into text, or text into audio, and it understands intent.
“Cognitive Services is about taking all of the machine learning and AI smarts that we have in this company and exposing them to developers through easy-to-use APIs [application programming interfaces], so that they don’t have to invent the technology themselves,” said Mike Seltzer, a principal researcher in the speech and dialogue research group at Microsoft’s research lab in Redmond.
Custom Speech Service makes use of an algorithm that shifts Microsoft’s existing speech recogniser to developer-supplied data. By starting from models that have been trained on massive troves of data, the amount of application-specific data required is greatly reduced. In cases where the developer’s data is insufficient, the recogniser falls back on the existing models, said Microsoft.
“The basic idea is that the more focused the systems can be, the better they will perform,” said Seltzer. “The job of the Custom Speech Service is to let you focus the system on the data that you care about.”
Amazon targets UK Alexa developers
Amazon is also expanding its developer support for Alexa, the online retailer’s AI-based voice recognition and speech service with availability of the Alexa Skills Kit in Europe. This will help developers create so-called “Alexa skills”, apps for Alexa. According to Amazon, European brands planning to integrate Alexa into their applications include JustEat, the BBC, The Guardian, Jamie Oliver, MyTaxi, Hive, Netatmo, National Rail and Deutsche Bahn.
John Rakowski, director of technology strategy at application monitoring firm AppDynamics, said: “Early iterations of Amazon Alexa pairings have included a voice-activated radio and in-car assistant, which gives some idea of its use cases. However, this is only the beginning. The fact it can be featured within other devices will allow users to give voice commands without having a dedicated device in every room. It also opens up the possibility of use in specialised sectors such as manufacturing, where niche solutions are needed.”
Industrial use of AI could lead to greater flexibility and customisation in manufacturing. For instance, an industrial company would be able to understand the reason a particular raw material is being purchased, which would then enable it to offer individualised product manufacturing.
Retailers such as Yoox Net-a-Porter Group are assessing how AI, using IBM Watson, could be deployed to make e-commerce more intuitive.
One of the AI application areas retailers are considering is the use of chatbots to reduce the volume of queries human customer care staff handle, allowing them to deal with more complex customer service enquiries.
Using AI for automation
Steve Hewitt, head of retail at Capgemini, said many retailers are looking at chatbots to provide more intuitive access to data. AI is also being used to automate metadata, which is important both for internal and external internet searches.
In the banking sector, AI is also being used for back office automation, customer service and fraud detection. Swedish bank SEB, for instance, used IPSoft’s Amelia AI engine to automate four IT processes: unlocking a user account, password guidance, unlocking a customer mortgage account and a more general guide to help internal users find an IT service.
In healthcare, DeepMinds, the artificial intelligence company owned by Google, is also expanding its AI work. At the end of 2016, the company announced it had begun work with Imperial College Healthcare NHS Trust to investigating the opportunity for mobile clinical applications to improve care.
Intel has also joined the efforts to make AI easier for developers to start using, with the introduction of its deep learning library, BigDL, which it claims will make AI training and tools broadly accessible to developers through the Intel Nervana AI Academy.
BigDL is a distributed deep learning library built on the Spark architecture. It makes deep learning more accessible to big data users and data scientists.
Developers can write deep learning applications as standard Spark programs that run on top of existing Spark or Hadoop clusters to put deep learning workloads more directly in touch with the data they use. BigDL is already running in the Databricks Spark Platform.
Unlike traditional programming, where a set of instructions are followed by the computer to achieve a particular objective, AI systems are taught using masses of data. When presented with new information, the AI uses pattern matching to make a best guess of what the data means.
AI represents a different way of thinking about computing, and the industry will need to work at getting developers to think differently about how to solve problems in AI. Amazon is holding an event in London on 15-16 February, focusing on Alexa. This is just the start, and the tech giants are beginning to look at tackling the knowledge barrier to entry.