Whole new world as robots begin to show their mettle
HOW close is artificial intelligence (AI) to changing how we see the world?
The technology recently made a grand entrance into mixed martial arts, in a way that could change how the story of a fight is told.
At re:Invent, an Amazon Web Services (AWS) conference in Las Vegas in December, mock sparring between two fighters used analytics and AI to measure everything from punch strength to fighters’ stress levels.
The vast data can be interpreted for audiences in real time, giving them an entirely new view of a fight. And why not, asks Mati Kochavi, one of the founders of the tech. “Shouldn’t sport be told in real time, with real data, details, insights, emotions?”
Also at re:Invent, online travel company Expedia shared how machine learning has improved the customer experience of its 600 million users.
CEO Mark Okerstrom says Expedia can perform 750 million searches, scan three million hotels in a second and provide 1.6 million customers with personalised results – every day.
When one considers applications of technology ranging from voice-powered personal assistants like Amazon’s Alexa, to autonomously powered self-driving vehicles with predictive capabilities, it is easier to imagine that AI could allow the machines to take over.
Computer-aided interpretation of medical images, heart-sound analysis and the composition of classical music by means of AI are just some of the advances in the field.
In his two-hour presentation, AWS CEO Andy Jassy unveiled 22 new products, drawing “aahs” and “oohs” from an audience enthralled by the powerful new capabilities in AI and its hi-tech siblings – machine learning and deep learning. Five that drew the loudest applause were:
● Amazon Translate: a “neural machine translation service that delivers fast, high-quality and affordable language translation”.
It uses deep learning to deliver more accurate and natural-sounding translation than older, rule-based translation algorithms, at a large scale, and in real time.
● Amazon SageMaker: “a fully managed service that enables data scientists and developers to quickly and easily build, train and deploy machine learning models at any scale”. It includes common machine learning algorithms like “off-theshelf” products.
● Amazon Rekognition Video: “a deep learning-powered video analysis tracker of people and activities that recognises objects, celebrities, and inappropriate content”. Called DeepLens, it uses neural networks to learn and make predictions.
It recognises faces in live-streams, analyses existing stored video, and automatically labels activities, people and objects, so the content can be searched. It’s expected to transform policing and public safety.
Matt Wood, GM of AI at AWS, says the camera can be programmed to do “almost anything you can imagine” – for instance, to recognise a car licence plate. So when you pull into your driveway, it can open the garage door automatically.
● Amazon Kinesis Video Streams: previously, the Kinesis product made possible real-time analytics using data streams.
The new version enables secure video streaming and other signals from connected devices for machine learning, analytics and processing.
It means, for example, that data from Lidar systems, which allow selfdriving vehicles to detect people and objects, can be processed faster and more accurately.
● Amazon Comprehend: a natural language-processor that uses machine learning to analyse large amounts of text. “It identifies the language of text, extracts key phrases, places, people, brands, or events, understands sentiment about products or services, and identifies the main topics from a library of documents.” This makes it useful for organising documents and analysing customer feedback.
The new products may seem like science fiction, but in five years will be regarded as the new norm in AI.
Wood says AI and machine learning, specifically robotics, lead to greater business value and allow businesses to grow at a faster rate. Closer to home
Wood says the top three industries that could benefit from AI in Africa are transport (autonomous cars), healthcare and manufacturing.
“We have many customers working on medical imaging, for example for early detection of tumours,” he says.
The manufacturing industry could improve the quality and output of its processes through automation.
The power of data
In the past, companies were constrained by the amount of data they could collect, how much computation they had access to or what hampered databases and analytics.
All that has changed with unlimited storage. Once those constraints start to fade, says Wood, the true value of the data comes into play. “You are able to move from not having any machine learning to very sophisticated machine learning in a week.”