Age of AI arrives
Artificial intelligence solutions are starting to make an impact on government, as organisations look to automate and enhance services and data systems
Artificial Intelligence – AI – is one of the most talked about developments in technology at the moment. From the creation of machines that are indistinguishable from humans, to the supposed threat of hundreds of thousands of jobs being taken by robots, AI is not only being talked about, but it is also becoming reality in the workplace.
After years of AI technology being showcased in competitions pitching human against machine in everything from chess to TV game shows, the technology is finally reaching ‘real world’ deployments.
There are many different definitions of AI, and a number of different technologies have been included in the category of AI. A good general definition is that Artificial Intelligence is when a computer can perform tasks that usually require ‘human intelligence’. As the technology has matured, the list of ‘human’ tasks which a computer can perform has grown, and the definition shifted with it.
The latest headline-grabbing achievement for AI came in January, when an AI system beat human players in a poker tournament, a significant achievement as the AI had to learn to make decisions without knowing all variables involved, and which also saw the AI learn the game on the fly.
In practical terms, AI technology is being brought to bear in several different forms. Capabilities of systems considered to be AI include parallel processing, to extract value from huge volumes of data; deep learning, to gain insight from sources unstructured data; intelligent automation, to understand and act on insights from data; cognitive analytics to learn from experience, and cognitive technologies which allow computers to interact and understand the world in ‘human’ terms such as speech recognition, object identification, or understanding natural language.
The real world applications of these technologies are reaching into new areas, across many different verticals, and many deployments are being developed for specific sectors. Analyst company IDC predicts that the leading uses of AI in 2017 will be in medical diagnostics and treatment, quality management in manufacturing, and automated service agents in retail. From 2017 to 2020, 70% of AI use cases will be industry-specific.
With such a broad range of capabilities on offer, the question for government is often where to start with AI?
Megha Kumar, Senior Research Manager, Software, IDC Middle East, Turkey & Africa, said that AI will be of interest within areas such as law enforcement, healthcare research and improved engagement with government agencies, although many of the initial adoptions are still at proof of concept or evaluation stage, and are only focused on narrow areas of technology.
“For the time being, government are clearly using cognitive systems rather than actual AI in the form of the engagement advisors,” Kumar said. “As these systems progress and learn to engage in proper decision making, will be able to see pure AI in play. In addition to this, you see government in the Middle East exploring the use of cognitive systems for call centre automation and augmentation, law enforcement and fraud detection as well as evaluating citizen sentiments.”
One area of initial uptake are these ‘engagement advisors’ - intelligent assistants that are used in decision support, often for customer support and engagement. Intelligent solutions such as ‘smart’ AI assistants, are helping to field enquiries and support customers in many different fields, and government is no exception. AI is seen as well suited to this field, where smart assistants can handle a large volume of enquiries, deploy natural language processing to understand enquiries, and machine learning to improve their responses. IDC predicts that these AI assistants will become commonplace, reaching over 110m consumer devices in the US alone by 2019.
Dubai Department of Economic Development is an early adopter in this area, launching it’s ‘Saad; service with Smart Dubai and IBM in October last year. Based on IBM’s Watson cognitive computing system, Saad allows users from the business community to ask questions and get up-to-date answers on business licensing and business registration in Dubai processes.
The Watson technology enables Saad to understand natural language enquiries, rapidly analyse and verify the relevant data from a variety of sources and then interpret the query and the data to provide relevant answers. Saad is available on the DED website and the DubaiNow app.
Other government entities which are using AI for customer interaction include Dubai Electricity and Water Authority, which launched a Facebook chatbot called Rammas, last month, and Dubai Customs has its Smart Virtual Agent.
Amr Refaat, general manager, IBM Middle East and Pakistan, explained: “IBM collaborated with The Smart Dubai Office, the Smart Dubai Government Establishment, and the Department of Economic Development to launch Dubai’s first cognitive computing service called Saad to facilitate business licensing and registration in Dubai. Saad helps position Dubai on the global stage as a smart and competitive destination for new businesses.”
Other areas where Watson is being used by government include healthcare, public safety and fraud detection, Refaat added. In healthcare, Watson is being used in the US by the Department of Veterans Affairs to assist doctors in delivering precision medicine – highly customised care plans - for cancer patients. Watson is able to provide precise information to physicians to treat 30 times as many patients as they could previously, and this level of data processing is proving its worth in many areas.
“We are seeing first-hand [cognitive computing’s] potential to transform businesses, governments and society. We have seen it turn big data from obstacle to opportunity, help physicians make early diagnoses for childhood disease, and suggest creative solutions for building smarter cities. We believe that this technology represents our best — perhaps our only — chance to help tackle some of the most enduring systemic issues facing our planet, from cancer and climate change to an increasingly complex global economy,” he said.
The new wave of AI is very much focused on, and enabled by, data. Big data sets, coupled with ease of access to the data and new algorithms to process the data are increasing the level of ‘intelligence;’ in AI, and AI is one of the few ways in which massive data sets can actually be processed in any meaningful way. Simply, many data sets are too big and too complex for human analysis alone, and only through AI support and machine learning can the value be extracted from the data, and fed back into the system for better decisions in future.
Data and AI is a core focus for the software vendors that are bringing solutions to government. Oracle has launched its Adaptive Intelligent Applications, which use cloud-based apps with deep analytics to process large volumes of customer data, and insight from Oracle’s own data cloud, to extract value from data sets. The apps are intended to make decision science and machine learning easily accessible to organisations, commented Si-Mohamed Said, Head of Marketing – ECEMEA, Applications, Oracle.
Microsoft is also working to make AI more accessible, from areas such as hardware field programmable gate arrays, which give servers flexibility in processing data, to AI assistants like Cortana, machine learning in applications such as translation services and making such applications available in the cloud.
Necip Ozyucel, cloud & enterprise business group lead, Microsoft Gulf, said that the company’s AI solutions are in use by a number of government customers worldwide, in a number of usage scenarios, including smart grids in San Diego, transportation in London, building and infrastructure management in Zaragoza and citizen services and tourism in Barcelona.
AI will be a key enabler, Ozyucel said, to scale government services, and also to manage large volumes of data. Microsoft’s solutions are intended to make the volume of data accessible, but at present, there is still a human requirement to have a data scientist to design any solution and ensure it is functioning properly, but from there, automation of intelligent actions will enable end-to-end processing without human interaction.
While autonomous systems and complete automation may be desirable, AI systems will need time to develop, and will require oversight if they are to be suitable for government, said Kumar.
“A lot of work will need to be done across educating cognitive systems to become responsive. You see intelligent assistants being a part of mobile ecosystem but they will have to go beyond such aggregating information especially if they are being used to access government services. This also will bring rise to the need for governance to be able to evaluate the decision making process of AI systems especially if they are being used to provision citizen services,” she said.