Why AI projects are failing?
To read the full article, learn about JOS Recommendation and receive updates, please register.
The reality gaps
Augmenting humans is complex. In many cases, the problem of an AI system lies in the comprehension and responses of an interaction.
For example, in chatbots the success lies on their ability to identify customer queries, which can often be ambiguous. Many chatbots tend to be very good at replying to customer queries using available FAQ, but struggle or give wrong responses when the queries differ in context or are asked in a different manner.
Equally important is the architecture that manages and feeds the data into the algorithm. Chatbots, unlike humans, cannot make cognitive leaps and use intuition to drive responses. Data availability and quality is vital for chatbots to offer relevant responses.
Yet, it is easier said than done according to Forrester. The research firm’s report Predictions 2019: Artificial Intelligence noted that “data doldrums will continue to drown the majority of firms embarking on AI”.
Realign AI, automation and productivity
Making AI real to business in 2019 will require business leaders to bridge these gaps and resetting expectations of AI.
With rising understanding of the technology, more are recognising AI’s impact in productivity.
A global survey by Economist Intelligence Unit (EIU) noted the top three areas that AI will bring a positive impact in their country/region and industry over the next five years is growth, productivity and innovation.
The recent 2018 Economist Corporate Network (ECN) study also shows 34.2% of Asia CEOs stated the top impact of AI is to increase productivity, compared to 23.8% in 2017. Some companies have started invested in AI for productivity gain.
“In a power company, AI is really improving many business processes,” said Vidal Fernández, Director Big Data & AI, China Light & Power. “One of them is saving cost in (power) generation process, so AI can optimise the generation process in our plants.”
“AI is definitely a core component and one of the key technologies we want to embrace and have embraced,” said Knattapisit Krutkrongchai, CMO, AIA Hong Kong and Macau.
The leading insurance company deployed a chatbot system at its contact centre in Hong Kong to drive productivity. “We apply the (chatbot) technologies to help staff in providing the most appropriate assistance to the customers,” added CTO Mark Seifried.
First step: Consider RPA
To realise the opportunities in AI to increase productivity, more businesses are turning towards automation technologies, specifically robotic process automation (RPA).
The momentum in RPA adoption has started way before AI. Businesses have been treating these technologies distinctly—RPA for automation, AI for intelligence. But increasingly, RPA is considered as a gateway for businesses to embark on their AI journey.
“RPA is laying the groundwork for machine learning and AI,” said Rory Yu, Head of Digital Solutions, JOS. “If customers want to start their AI journeys, automation is key.”
RPA uses machine learning to automate business workflows, and mimic human interactions with systems, offers an ideal platform to realise the benefit of AI.
One reason RPA is considered as a starting point is its relatively easy strategy into automation of back-office processes. It is particularly well suited to working across multiple back-end systems without re-architecting those systems.
RPA also brings a quick and high return on investment in early pilots. A study from American Productivity and Quality Center found that over 75% of respondents said their early RPA projects had met or exceeded expectations. Meeting expectations may be easier for RPA, given it often has a clear process to automate and a measurable business case.
Additionally, RPA reduces manual workload and allowing staff to focus on value-added tasks or those that need human input.
This is why many financial, education, manufacturing, logistics and legal companies are turning to RPA to start their AI journeys, driving growth in the RPA market. The Everest Group expects the RPA software market to grow between 75%-90% in 2019.
While AI offers a lot more, RPA can allow companies to quickly understand the value by offering measurable benefits in shorter time.
“RPA allows phases of implementation starting with simple bots that gather the input and output of data, then progressing to cognitive bots that make decisions or interact with humans involving machine learning,” added Yu.
“RPA and AI will join forces to create digital workers for more than 40% of enterprises,” according to Forrester report Predictions 2019: Artificial Intelligence. The research firm further stated that “an RPA plus-AI technology innovation chain will turbocharge your innovation efforts.”
The human factor in RPA
While business leaders and decision makers love automation, employees see it with a wary eye. Reason: job security.
Many see automation initiatives as replacing themselves, which can be true to a certain extent. But humans are still needed to ensure that the AI engine is learning correctly.
Human domain experts of business knowhow within the organisation is equally valuable as technical expert from external partners. The main reason is that AI algorithms do not differentiate correct or incorrect data, bias or cultural nuances at the beginning. A human “expert” can.
Hence, companies need to build a corporate culture that allow employees to see automation as an advantage to their jobs, offloading manual tasks and streamlining other processes. Communication is key in building such culture and transforming through disruptions. To shape and build this future workplace in the organisation successfully, business leaders need to make sure the entire organisation get the message.
“Without a clear strategy across the workforce, businesses will struggle to drive value from their RPA initiative,” added Yu.
Big bang theory not advised
RPA can be disruptive to current company processes and culture. Companies looking to embrace it should begin with small projects, which offer a two-way platform for learning: companies can hone their processes to make RPA successful, while employees can learn to maximise their benefits and address their fears of displacement.
Companies are advice to start with the low hanging fruits where automation and cognitive technology can make an immediate impact.
Automating internal and small interdepartmental processes are good areas to start. Ultimately, AI projects require data and processes across the enterprise, they need different departments to work closely, which is never easy to accomplish. Thus, small interdepartmental RPA projects bring a head start for departments to work with each other and experience the benefits of collaborative creation.
AI is an area that is evolving, but in a talent-starved economy getting the right skillsets and knowledge can be tough. Having the right partner with constantly updated knowledge is important to traverse the reality gap.
More importantly, companies need to start their AI journeys right now. Else, they risk being cognitively outmanoeuvred in the near future.
How do you feel about embarking on an AI journey in 2019?
How do you feel about bringing RPA to your organisation?