By Sayanti Bhattacharya
Remember the Japanese robot hotel that opened in March 2015? The Henn na Hotel in Nagasaki, Japan, made its way into the Guinness Book of World Records as the first robot-staffed hotel. It demonstrated the power of AI in supplementing technological evolution. However, in less than four years of operation, the hotel made headlines again. This time by firing half of its robotic staff and returning to human-serviced operations.
AI/ML enthusiasts will wonder why this initiative failed since the use case is well-suited for scaling operations. The reason for rethinking this servicing model mainly came down to consumer satisfaction. Guests complained that the robots would break down, misinterpret human sounds (like snoring) as commands, the front desk robot receptionist struggled to answer guests’ custom questions etc.
Companies are inclined towards AI/ML solutions to drive business strategies at scale. However, a key question to ask would be while such solutions bring efficiencies in the process, do they maintain (if not enhance) the consumer experience.
The hospitality industry’s key objective is to provide consumers with an enjoyable experience which is subjective and needs human intervention. The consumer needs were significantly overlooked in the AI application.
Another example is Galactica AI launched by Meta in November 2022, which was taken down in three days. Galactica was introduced as a large language model, an enabler for researchers and students to research, summarise and generate scientific content.
The reason for Galactica’s short-lived stardom was its inability to detect truth from falsehood.
AI/ML has extremely valid applications and can even save lives (Apple watch alerting on possible heart attack, Tesla autopilot averting accidents etc.) but they must be developed keeping the consumer experience in mind. Machines can simulate human behaviour but with today’s complex societal structures, human responses are getting fuzzy and deviant. This requires more focused and personalised AI/ML.
In such a context, while an algorithm may be able to do its part in addressing a problem, its developer must do due diligence by defining it. One must overcome assumptions and bias on data and behaviour, be flexible with approaches, identify and tighten model accuracy parameters and plan for outliers.
To conclude, while implementing an AI/ML approach, it is critical to identify how it will impact the end user, then leverage enough data (including fringe cases) to capture consumer needs. We have some way to go before achieving these and it may be better to leverage a human-augmented AI/ML approach until then. This is what Hennna Hotel did — they deployed robots in areas like taking food orders, where they were efficient, and continued to leverage the human connect to deliver an enjoyable consumer experience.
The author is associate vice-president, Merkle