August 01, 2018

Can AI take customer service to the next level?

Written by 
  • Print
  • Email

MENTION Artificial Intelligence (AI) and it’s difficult not to picture scenes of biorobotic androids clashing with their human creators. While the reality of AI in 2018 is rather less dystopian than the Los Angeles of 2019 in the sci-fi classic Blade Runner, there’s no shortage of hype around the potential impacts, both positive and negative.

While it is difficult to determine exactly how fast and how far AI will go in terms of disrupting the business value chain, there is a consensus that the ability to automate processes, analyse data beyond human comprehension, and personalise customer services with unprecedented precision will have profound and far-reaching impacts on how companies operate. The economic effects of such changes are likely to be of equal significance - according to analysis by PwC, AI could contribute up to €13.3bn to the global economy by 2030, including €5.6 trillion from increased productivity and €7.7 trillion from opportunities relating to customer experience.

The potential applications of AI and its impact on rail operators and passengers was a key theme at last month’s Singapore International Transport Conference and Exhibition, hosted by the International Union of Public Transport (UITP) and Singapore Land Transport Authority (LTA).

AI Aug18The UITP Asia-Pacific Centre for Transport Excellence is nearing the conclusion of a research project into the role of AI in public transport systems and striking an effective balance between human interactions and technology. The Artificial Intelligence in Mass Public Transport project included a qualitative survey of attitudes to AI, which received responses from more than 50 public transport stakeholders. This found that 31% of companies surveyed are currently using or offering AI technologies, and a further 31% are testing or developing AI-based solutions. Most began investing in AI less than three years ago.

The study also includes 17 detailed case studies of AI applications in public transport. According to Mrs Gayang Ho, research policy development lead for the Centre for Transport Excellence, successful early adopters of AI have targeted “low-hanging fruit” - simple projects that quickly start generating tangible results.

Chatbots - AI conversational agents - are a prime example of this, and are being widely adopted by operators around the globe in a bid to enhance customer services. According to Ho, advantages include low psychological barriers and ease-of-use; ease of access to ready-made solutions, which can be easily integrated into existing platforms; and relatively low technology investment, with hardware and technology costs ranging from €70,000 to €110,000.

However like all AI technologies, Chatbots require a learning curve, and initial errors or limitations of the technology can drive users away. First impressions are important, and the conversation with the user needs to be “smart and goal-orientated.” A robust stress test is required to ensure the Chatbot can cope with a sudden influx of enquiries during service disruptions.

Chatbots have been deployed by Transport for London, which has integrated its TfL Travelbot into Facebook Messenger, and MTR, Hong Kong, which has integrated a chatbot into its mobile app.

RATP Dev, France, has used machine learning to improve customer analytics. The operator’s Interstellar AI engine, which was initially deployed on tram networks in Valenciennes, France, and Casablanca, Morocco, draws on operational, ticketing, timetable and external data to gain deep insights into the passenger experience that would not be possible through traditional (and costly) origin-destination surveys.

This has enabled RATP Dev to adopt a data-centric approach to resolving congestion issues and analyse different scenarios through simulation and prediction during disruption events. RATP Dev says Interstellar has cut contractual penalties by reducing the risk of delays, with better operating and planning, and enabled it to increase peak capacity on the Casablanca light rail network by 10% with the same number of trams in operation.

East Japan Railway (JR East) is using digital assistants with Natural Language Processing (NLP) and pattern recognition to improve service quality. These include the deployment of multilingual customer service robots such as Hitachi’s Emiew3 (pictured) at busy stations in Tokyo and a call centre support solution, which is based on IBM’s Watson AI platform. JR East service information can also be accessed via Amazon’s Alexa virtual assistant.

However, speakers also stressed that any AI strategy needs to carefully consider the limitations of deploying such technologies in customer service functions. “In the overall system there has to be experience, knowledge and also emotions,” explains Mr Alok Juin, managing director of Trans-Consult Asia and a consultant on the UITP AI study. “A machine would never allow you to put the life of a human at risk to save the life of another human. Humans do this all the time because they have emotions. Staff have delivered babies in stations, a machine will never be able to cope with such a situation. So AI is a valuable tool, but it will never completely replace humans on the railway.”

Get the latest rail news

IRJ Rail Brief newsletter covers global railway news