Driving digital transformation with new and emerging technologies.

AI is the future – but the cloud can get you there right now

Written by Tech editor of nextmedia Pty Ltd

Australian businesses and government agencies are enthusiastic about the potential for digital transformation to improve efficiency, increase profits, and serve customers better. Yet even as innovative technologies like big data, artificial intelligence (AI), and machine learning (ML) emerge to facilitate digital transformation, many organisations risk being left behind because they lack the resources to successfully harness those technologies for their own purposes.

The need to be more responsive to customer histories and behaviour has driven many companies to tie their digital-transformation strategies to new AI and ML technologies. These tools offer new ways of processing large volumes of events as they happen, helping organisations engage with customers and end users in increasingly intuitive ways. And, with cloud providers now offering instant access to powerful AI engines trained on billions of data points, it’s easier than ever for developers to tap into AI to enhance the applications that are driving their organisations’ digital transformations.

AI is already paying off. “Our customers embrace the cloud because they know we can make a difference in how they interact with technology,” Rodney Haywood, head of architecture with Amazon Web Services ANZ, told the audience at the recent AWS Leadership Forum in Melbourne.

“They want their business to be digital, but it’s not only about being digital: they want to harness the power of the cloud to make a real difference to their business at a tactical and strategic level.”

Organisations that have been proactive in embracing AI have already seen significant benefits. The recent Capgemini State of AI survey found that around three-quarters of the organisations implementing AI increased sales by more than 10 percent, improved operational efficiency by more than 10 percent, and enhanced customer satisfaction by more than 10 percent.

Each of those outcomes translates into real business value, making it obvious to any executive why implementing an AI strategy is important sooner rather than later. Yet not everyone is rushing towards AI and a recent Morar Consulting-Epicor user survey found that many still don’t even understand it.

Thirty-two percent of surveyed Australian business leaders admitted being unfamiliar with machine learning, with 36 percent unaware of cloud application-as-a-service models and 30 percent unfamiliar with the idea or benefits of big data.

“If you went back two years ago and tried to create that type of machine learning AI, image recognition, image classifications system, you would have needed data scientists, machine learning engineers, and a huge amount of learning and expertise to build that up, Domain did that for $523,”” Gore said.

AWS is continually developing its family of AI services in other ways: Amazon Lex, for example, encapsulates automatic speech recognition and natural language understanding that lets users interact naturally with information services. Amazon Polly delivers lifelike natural-language speech in more than two dozen languages.

Cloud-based access to AI has saved Carsales, Domain, and many other companies from the complex and burdensome task of building their own technology, instead allowing them to focus on innovative new ways that AI can improve their businesses.

AWS’s AI services and ML platform accommodates nearly any kind of business, enabling digital transformation even for those that don’t have the resources to build their own specialised AI research and implementation teams. AWS offers Amazon Machine Learning and Apache Spark on Amazon EMR and for developers looking for infrastructure on which to develop intelligent applications, AWS provides Deep Learning AMIs, a secure and scalable environment for Deep Learning on Amazon Elastic Compute Cloud (EC2).

Bridging the skills gap. Over time, the developer community is learning more about AI and how to apply it to business problems. A growing body of case studies is testament to the benefits that can be achieved, with AI proving itself to be a powerful enabler for digital transformation.

“In Australia we have built a huge amount of IP, expertise, and experience in how to refactor applications moving from on-premises to the cloud,” Gore said. “We know how to build digital native applications, and how to be experts in DevOps and microservices. And that’s what’s game-changing about bringing AI to the cloud: it allows us to use world’s-best technologies that have been exercised billions of times per day.”

Many of these technologies have direct implications on the quality of customer service delivered by an organisation. Pairing AI-based speech recognition and cognitive processing with an extensive database of support information, in particular, has significant implications for improving and automating customer-support processes that have traditionally been limited by the number of humans available at any one time.

A company could, for example, harness the capabilities of Amazon Lex and Amazon Polly to build a fully interactive chatbot that could serve as a first-line interface with customers – automatically handling routine tasks like password changes, account balance enquiries, and status updates without requiring human intervention. In fact, what’s perhaps most intriguing is that customers are already beginning to come up with combinations and end uses with AWS AI that we didn’t necessarily predict – and they are trialling them to solve real world problems.

These three technologies together constitute major drivers for transformation, and their use was correlated with business performance: 88 percent of fast-growing businesses in the survey consider IT investment to be a high priority, compared with just 41 percent of businesses that are experiencing weak growth.

The AI skills gap. Even among those companies that recognise the importance of transformative technologies, actually applying them is proving challenging.

But the evidence suggests that they should persist because the game-changing benefit of cloud-based AI is clear. AWS is democratising AI and offering every company the opportunity to take advantage of the innovation being delivered by its thousands of specialist engineers. Big data, AI and ML technologies, in particular, are rooted in complex mathematics and algorithmic science skills – and these cannot, AWS Chief Architect Glenn Gore warned, be easily developed inhouse.

“Never before has there been so much awesome technology, driving such extensive innovation and change. You now have access to technologies you could only have dreamed about a few years ago. And the great thing about ML is that the more you use it, the smarter it gets. But with so much change happening, the question is how we can keep up with it,” Gore said.

Effectively integrating AI and ML skills into an enterprise requires a combination of business, development, and data-analytics skills that few workers possess – and few organisations have ready to go.

“Many people think it will just happen by osmosis, and that their teams will pick up new technologies over time. But Machine Learning is not in this category. If you want to dive deeply into it, it takes a conscious effort in terms of what skills you need, and how you want to apply those within the business,” Gore said.

Figures from IT recruiter Hays suggest that many companies are looking outside the organisation to keep up with the skills they need to support their digital transformations with 37 percent of IT departments confirming they would be increasing their use of temporaries and contractors this year, particularly high-skilled staff that bring specific skills such as analytics and AI.

Yet there is a limited supply of such specialists, and companies can’t expect to just bring in an expert and have all their problems solved. With the Epicor figures suggesting that many companies are still getting their heads around the idea of AI and ML, many Australian companies just aren’t ready to take advantage of the transformative technologies that they need to remain competitive into the future. Those that fail to find a way to embrace them, will soon find themselves outpaced and out-innovated.

Rolling along with AI. The experience of online car-listing provider Carsales highlights how transformative AI can be. With 9 million unique visitors launching 800 million searches per month, the business has “a lot of plumbing under the hood”, Michael Ridgway, Carsales’ development manager and director of AI & ML, told attendees at the AWS Leadership Forum.

“We thought innovation within carsales was in a healthy state – but we looked at the numbers and realised that if innovation was going to keep moving at a fast rate, how were we going to adopt that to reduce our time to market, speed up things, and do more with less?”

That realisation stimulated an overhaul of the way the company operated, transforming it from an inwardly-scaling business that was producing one new software release per month into an agile, distributed business that now produces up to 300 new releases per month.

Making this transformation required both organisational transformation – Carsales split its monolithic business and development teams into 23 different groups each handling different parts of the Carsales site’s functionality – and technological advancement, which was delivered by moving the business to the AWS Cloud.

Carsales originally moved to AWS to provide scalability and reliability for its continued growth, but the team quickly recognised that AWS was investing extensively in innovation around AI and ML. Its large-scale AI engines – which are being continually refined and improved based on billions of real-world data points – could be integrated with Carsales’ applications within minutes using standard API calls.

“We had a lot of smart people but we weren’t really empowering them to the extent that we could,” Ridgway said. “We decided to empower our teams with whatever they needed to deliver a higher-quality product.”

Carsales encouraged staff to explore new ideas using a series of hackathons and the promise to put at least half of the ideas they produced into production. Staff began experimenting with AI capabilities that Carsales used to develop systems such as a photo-recognition tool called Cyclops.

Cyclops lets Carsales address a specific challenge that vendors faced when uploading photos of their cars, many users would forget to include photos of key elements such as the boot, dashboard, engine, or both sides of the vehicle.

By leveraging AWS’ AI capabilities, Carsales staff were able to teach Cyclops how to recognise the many different parts of a car, and automatically label them. The system has proven to be much more accurate than having humans do the same task: in Carsales’ testing, Cyclops accurately labelled photos 97 percent of the time, compared with the 85 percent accuracy of a human control group. It also works more quickly and is available virtually anytime.

Cyclops also enables better comparisons between cars. For example, by showing photos of the boot space of several cars of interest. This improves the quality of listings for vendors and the depth of information available to buyers, which provides critical competitive advantage.

“We are starting to uncover all kinds of interesting metadata in the images, and we’re using that in new ways,” Ridgway said. “The transformation journey is one of those things that you start and you keep doing it and doing it. We want to keep improving every single day – because every day we don’t make a change is a day that our competitors do.”

The cloud AI advantage. Carsales’ move to the AWS cloud gave it more scalable, flexible infrastructure. Just as importantly, the cloud also provided its developers access to advanced AI capabilities that they could immediately leverage for competitive advantage.

Real-estate listing giant Domain Group is also tapping into AWS AI to explore innovative new services. The company has leveraged computer-vision tools for an image-recognition project enabling automatic analysis of property photos, allowing users to search for all houses with specific features, such as white kitchens, brick garages, or outside BBQs.

Building such a model would have taken Domain’s engineers a massive amount of time and money. However, by feeding 210,000 training images into AWS Rekognition, they were able to get the same result for just $523 worth of cloud-computing resources.

“Customers are already beginning to come up with combinations and end uses with AWS AI which we didn’t necessarily predict and they are trialling them to solve real world problems,” Gore said.

Rapid commercialisation of those services will make AI-optimised customer service and back-end services an everyday part of life in 2018 and beyond. A growing base of data, generated by ever-larger networks of sensors and other Internet of Things (IoT) devices, will provide additional data to help ML models make personalised, contextually-relevant services to enhance the customer experience.

Because such capabilities are delivered from the cloud, they’re fully scalable and ready to be routed to whatever device the customer is using at the time – helping companies deliver consistent, effective support that improves the overall customer experience.

“Developers don’t want to think about servers,” Olivier Klein, head of emerging technologies with AWS Asia Pacific, told the AWS Leadership Forum. “Developers just want to be where the customer is, and to make the customer’s experience as naturally engaging as possible. At the same time, we can’t waste massive amounts of effort doing this manually. AI and chatbots will help us do that automatically and efficiently, making better decisions about personalising the customer experience.”

With so much data being fed into the mix, ML models will pave the way towards more rapid innovation as developers look to surface the insights they provide in different ways.

Augmented reality, for example, can overlay critical sensor or status information on top of industrial equipment, and use visual queues to assist in diagnosis and maintenance tasks. Travel apps can overlay intelligently-derived route information based on historical analysis of delay trends and slowdowns due to road works. Or an app could allow the user to point a smartphone at a foreign-language speaker, feeding the audio into Amazon Lex, and overlaying the translation onto the video like subtitles in a movie.

“Moving forward, we’re going to think differently about how we interact with consumers,” Klein explained. “Everything will connect and that’s very powerful. And we need to have the ability for applications to actually adapt for humans to use them in a natural, personalised manner.”

The possibilities are endless and the need to act now is real. By leveraging the expertise of thousands of developers and millions of person-hours, developers can kickstart innovation and leapfrog the competition by introducing new consumer-facing features that refine the customer experience – and propel the entire company towards finally realising the possibilities of digital transformation.

 

 

 

Preparing for an AI future

Machine learning and artificial intelligence technologies.

Learn More
In the News

AusPost trials machine learning to manage unpaid bills

Australia Post has quietly created an email add-on tool that uses machine learning to find unpaid bills and itemise when they need to be paid.

Learn More

Get in Touch with Us

Whether you are a start-up, mid-size or enterprise sized business, let us know how we can help you with actionable insights and pave a path forward to a successful digital transformation.

Contact us

No posts available
for this selection.

Please filter again, or search below.