ROBOTICS & ARTIFICIAL INTELLIGENCE. White Paper

Transcription

1 ROBOTICS & ARTIFICIAL INTELLIGENCE White Paper

2 CONTENTS ROBOTICS AND ARTIFICIAL INTELLIGENCE THE PANEL WHAT DOES THE ROBOTICS AND ARTIFICIAL INTELLIGENCE INDUSTRY LOOK LIKE TODAY? WHAT ARE THE CHALLENGES OF BRINGING ARTIFICIAL INTELLIGENCE INTO THE CORPORATE SETTING? WHAT DOES THE FUTURE LOOK LIKE FOR THE ROBOTS AND AI INDUSTRY? Robotics and Artificial Intelligence White Paper

3 01 ROBOTICS AND ARTIFICIAL INTELLIGENCE There has been an explosion of interest in the field of robotics and Artificial Intelligence (AI) over the past few years, as companies begin to recognise the potential for technology to revolutionise the way they do business. There has also been much hype about the advance of the robots, with research from the World Economic Forum suggesting that as many as 15m jobs will be replaced by robots over the next 20 years. While it s unlikely that robots will wipe out humans any time soon, the reality is that AI is already alive and well, and being used widely in Australian industry. Advances in technology have had a massive impact on everything from the way companies design and make products, to the way they deliver their services and interact with customers. Of course, this also has huge implications for the way work is organised, the way people are managed, and the type of skills that will be needed in the future. We are seeing new roles emerging, with studies estimating that 65 per cent of children currently at school will end up doing jobs that haven t yet been invented. Morgan McKinley recently brought together a panel of Australia s leading experts in robotics and AI to talk about what is really happening in this exciting field, to discuss the challenges of bringing AI into the corporate sector, and to provide some insight into what the future might hold. We are truly on the cusp of the next industrial revolution, with companies in fields ranging from healthcare and manufacturing, to automotive and financial services, already employing AI to streamline their processes and improve the service they offer to customers. Robotics and Artificial Intelligence White Paper 01

4 02 THE PANEL DR CATRIONA WALLACE CEO & Founder at Flamingo AI Dr Wallace is the founder and CEO of FinTech and ASX-listed company Flamingo, which provides an Intelligent Guided Selling Platform (AI) and Cognitive Virtual Assistant called Rosie that helps financial services companies improve online sales conversion rates. ROBERT FEENEY Partner at A.T. Kearney Robert Feeney is a partner and leads A.T. Kearney s Strategic IT Practice for Australia and New Zealand. He specialises in developing and aligning technology and business strategies and partnering with clients through IT transformation. WILLIAM JUDGE Head of Social Robotics at Commonwealth Bank of Australia William Judge is the Head of Social Robotics at the Commonwealth Bank s Sydney Innovation Lab. He spearheads the bank s research and development programme into human robot interaction, which aims to explore the commercial applications of social robots BROCK DOUGLAS IBM Watson Leader Asia Pacific at IBM Brock Douglas is IBM Watson Leader Asia Pacific at IBM. He works with clients to promote positive brand experiences using cognitive tools such as artificial intelligence, robotics and analytics. Brock has led transformational projects for iconic brands such as David Jones in Australia and Lotte in Korea. Robotics and Artificial Intelligence White Paper 02

5 03 WHAT DOES THE ROBOTICS AND ARTIFICIAL INTELLIGENCE INDUSTRY LOOK LIKE TODAY? Many of us have already brushed shoulders with AI in the shape of Amazon s Alexa or Apple s Siri virtual personal assistants who respond to voice commands, and can help us do everything from putting together a music playlist, to dimming the lights, and picking up the latest traffic report. The same platform is also being used by online bank UBank, which has introduced an AI-powered chatbot to help customers fill out home loan forms. Other systems, such as Flamingo s Rosie, are being used to help businesses in the financial sector boost sales by providing an end-to-end, automated customer experience. These AI applications are, however, just the beginning, with a growing number of examples of sophisticated systems being used across industries. It s not unusual now for a chatbot to be processing customer orders or responding to basic customer enquiries, while in Japan; robots are already supporting the care of dementia patients in care homes. Although there is a long way to go, we are also now starting to see the emergence of cognitive systems that can understand people s intent and are able to reason, think, learn and interact with customers. One example of this is IBM s Watson platform, which is being used by office supply company Staples to streamline the ordering experience for its customers. Based on a simple voice command (such as; I want to buy some blue pens ), , text, or even photo of a product, Watson can use its bank of knowledge to work out exactly what type of blue pen the customer needs, based on information such as which brand they have ordered in the past, and how many they are likely to want now and in the future, all from order history. Robotics and Artificial Intelligence White Paper 03

6 03 WHAT DOES THE ROBOTICS AND ARTIFICIAL INTELLIGENCE INDUSTRY LOOK LIKE TODAY? Panel members described the three main types of Artificial Intelligence currently either under development or being actively used in the business world: ARTIFICIAL SUPER-INTELLIGENCE: Work is already being done on the development of super computers with intelligence levels that surpass the capabilities of the human brain. They have massive memories, unlimited analytical capabilities and the ability to multi-task in a way that is not biologically possible for humans. Developed to their fullest potential, these systems will be able to invent or discover almost anything. It may all sound a bit sci-fi, but experts predict we could see these kind of systems mature in our life-time. NARROW ARTIFICIAL INTELLIGENCE: We now have intelligent learning machines which can be trained to manage processes and interact with customers in highly specialist fields. So an insurance company, for example, can programme a machine to take a customer through every step of purchasing one of its retirement products. The computer can be taught the range of questions a customer will typically ask as they go through the journey of researching the product, getting a quotation, signing up to the product and making a payment. These chatbots are already being used within the financial services industry and are an affordable and easy to set up solution for companies working in specialist markets. GENERAL ARTIFICIAL INTELLIGENCE APPLICATIONS: The most common examples of these are the virtual personal assistants which can be programmed with a specific skill set and have the ability to respond to voice commands. As anyone who has used one of these intelligent systems will tell you, while they have a useful role to play, they don t necessarily get it right all of the time. The beauty of this kind of AI, however, is that robots and chatbots can be programmed to learn from each customer interaction, working out what they got right or wrong, and storing that intelligence into their knowledge base, ready for the next customer interaction. Robotics and Artificial Intelligence White Paper 04

7 04 WHAT ARE THE CHALLENGES OF BRINGING ARTIFICIAL INTELLIGENCE INTO THE CORPORATE SETTING? Artificial intelligence has the potential to disrupt business models and revolutionise the way companies interact with their customers. It can streamline processes, put information at customer s fingertips and free staff up from repetitive administrative tasks so that they can concentrate on more value-adding activities. Introducing it into a corporate setting, however, is not without its challenges. Drawing on their expertise of working with clients across industry sectors, panel members identified the following key issues. AI-READINESS The reality is that although organisations are excited by the idea of Artificial Intelligence, when it comes to the reality of introducing it into their operations, they are often not ready to take the plunge. They are reluctant, for example, to give vendors access to their legacy systems, either because they have heard stories about negative experiences or because they are concerned about handing over sensitive customer data. Data security and privacy is also top of mind for suppliers, many of whom have invested heavily in building special privacy features into their products. Some platforms, for example, are able to collect identifiable customer information, input it into legacy systems, then forget it. Suppliers are finding there are many businesses who are keen to have conversations with them about the potential of AI, but far less who are actually willing to take the first step on the journey. The bigger and more mature the company, the less likely they are to be AI-ready, providing a huge opportunity for smaller second and third tier businesses, who could potentially overtake their larger competitors in the race to improve the customer experience. THE BUSINESS CASE: Companies are often so excited about the potential of AI that they get caught up in the hype and overlook the need to build a strong business case. They know that there s really cool stuff out there and are keen to get involved, but haven t given sufficient thought to exactly why they want to embrace AI and what overall objective they are trying to achieve. Panellists felt that companies needed to have a clear strategy in place at the outset, and to have thought about what kind of applications they are likely to need, what sort of interface would work best with their customers and how they might exploit the wealth of data the systems can produce. Robotics and Artificial Intelligence White Paper 05

8 04 WHAT ARE THE CHALLENGES OF BRINGING ARTIFICIAL INTELLIGENCE INTO THE CORPORATE SETTING? SOCIAL ACCEPTANCE: There has already been some experimentation in the retail and leisure space with human-robot interaction. A major shopping outlet in Sydney, for example, recently hosted a humanoid robot ( Chip ) who helped to guide people around the centre and distributed product samples to shoppers. Although there is great potential for this technology, developers still face a number of challenges. There are technical issues with speech recognition, for example, especially against noisy backgrounds such as hotel lobbies or retail outlets. Robots also have limited battery life and rely on a good Wi-Fi signal to operate efficiently. Perhaps the biggest challenge, however, is developing robots that understand social norms and are able to interact on an emotional level. More work needs to be done on developing technology that will enable robots to recognise facial expressions, so that they can pick up on customer emotions and respond accordingly. A CONSERVATIVE APPROACH: Companies are often reluctant to put their AI technologies out into the market for use by customers and employees until they have been rigorously tried and tested. Although it may sound counter-intuitive, this conservative approach actually hampers progress. Robots and chatbots learn best in an end user environment where they are having regular interaction with customers. The user feedback enables them to build up a bank of knowledge so that they are more quickly able to meet customers needs. It calls for the organisation to make a leap of faith As an example; Ubank put its chatbot live with customers after only eight weeks of development but the end results are worthwhile. There is also some debate about the way robots are named. Research has suggested that companies believe robots that take on young, white, female personas are generally the most acceptable to the public, although similar research among consumers puts gender neutral robots at the top of the list. This clearly doesn t reflect diversity in society and is an area which needs further attention. Robots are unlikely to achieve widespread acceptance in public settings until these challenges have been met. Robotics and Artificial Intelligence White Paper 06

9 05 WHAT DOES THE FUTURE LOOK LIKE FOR THE ROBOTS AND AI INDUSTRY? Panellists acknowledged that although this was a hugely exciting industry with many positive opportunities, there was also a potential dark side to AI that needed to be carefully managed. A number of governance bodies have already been set up in Australia to regulate the industry, and measures are being put in place to avoid the risks of centralisation of AI in the wrong hands. Thanks to much of the hype and media coverage around robotics, there is currently a significant gap between expectations of what AI can deliver and the on-the-ground reality. Technological development is, however, continuing apace with more sophisticated applications coming onto the market at a rapid rate. Some of the future trends we are likely to see in this space include: ROBOT-HUMAN INTERACTION: There are significant opportunities for robotics to be used in healthcare and education. In both Asia and Europe, experiments are taking place in elder-care, where robots are being used to support people continuing to live independently in their own homes. The robot can bring the person water, remind them to take their medication, identify if they have had a fall, and initiate contact with emergency services if needed. Work is also being done with dementia patients in care homes, where robots have been programmed to interact with residents and remind them of stories from their past, or songs which are relevant and meaningful to them. This provides much-needed social interaction and reassurance and helps to ease the pressure on staff. Research is also being conducted into the potential for robots to work in educational settings with pupils who are autistic and struggle with social interaction. A robot can be programmed, for example, to strip away some of the layers of non-verbal communication that people on the autistic spectrum often struggle with. These levels of interaction can be gradually re-introduced, one at a time, making it easier for pupils to interact with others at their own pace. One of the challenges of developing technology for use with vulnerable people is making sure there are no opportunities for emotional manipulation, be it intentional or unintentional. The challenge is that people can build strong bonds with these human-like robots, which can have a negative as well as a positive impact. Robotics and Artificial Intelligence White Paper 07

10 05 WHAT DOES THE FUTURE LOOK LIKE FOR THE ROBOTS AND AI INDUSTRY? FINDING THE RIGHT MODEL FOR DESIGN: Finding the right human model to replicate is one of the issues designers in the robotics field are grappling with. Everyone has different views on how people should think, relate and interact with others and finding the right norm has already proved to be a challenge. One chatbot, for example, was unleashed onto Twitter and had to be quickly withdrawn when it started to develop some very undesirable xenophobic, racist and homophobic tendencies all of which it had learnt from fellow Twitter users. Parents have also complained about the impact virtual personal assistants such as Alexa have had on their children s behaviour, encouraging them to make demands and forget some of the common courtesies such as please and thank you. It is unlikely cognitive systems will ever be capable of developing morals so careful attention needs to be paid to exactly what model of human behaviour their design is based on. There was some discussion about whether robots should be created in the human mould at all, especially when they were being developed to manage tasks that humans are not typically good at. IMAGERY: Exciting work is taking place in the area of image recognition. In the medical field, for example, work is taking place on the early detection of melanomas. Computers can scan images of people s moles, determine whether they are high risk and refer the patient to a doctor with a recommendation for appropriate treatment. Some utilities companies are using satellite imagery to identify leaks in underground water pipes, whilst in food manufacturing; imaging technology is being employed to check for packaging issues. More sophisticated use of this technology is probably 18 months to two years away, but the potential is enormous. Robotics and Artificial Intelligence White Paper 08

11 05 WHAT DOES THE FUTURE LOOK LIKE FOR THE ROBOTS AND AI INDUSTRY? OLD JOBS VERSUS NEW JOBS: Much concern has been expressed about the potential for AI to eradicate jobs. There is no doubt that this will happen in certain sectors and with certain roles. In a typical call centre environment, for example, companies have the potential to make efficiencies that could lead to a 30 per cent reduction in their workforce. However, while old jobs will undoubtedly disappear, new jobs will emerge. These will not necessarily be high level data scientist jobs, but a new type of support role. Preseeders, for example, will be needed to train robots and provide the input and information they need before they can start to interact with customers. We are also likely to see the emergence of human virtual assistants, who will work alongside the robots to pick up questions that the system has not yet been programmed to answer. Data journalists people who can interpret data and turn it into stories that will help corporates improve the customer experience on an ongoing basis will also be at a premium. The challenge for organisations will be whether they are able to re-skill employees fast enough to keep up with the pace of change. CHANGES IN THE VENDOR MODEL: Rapid developments in technology will also lead to changes in the pricing and delivery of services. Vendors are likely to move to a new model where charges are based on the number of connections or interactions a system has with customers, as opposed to the traditional licence fee. Panellists also predicted a shift to more collaboration between vendors, who will increasingly work together to develop eco-systems of solutions that are modulised and networked to meet the full range of customer needs. This move towards greater partnership also reflects the way robots and humans need to work together in the future. As the panel concluded, there are some things that robots can do better than humans, and some aspects of human behaviour that robots will never be able to replicate. We need to work towards an ideal scenario where both work together and complement each other. Robotics and Artificial Intelligence White Paper 09

12 morganmckinley.com.au