Customising AI strategies through experiments

ARTIFICIAL INTELLIGENCE (AI) is not being implemented optimally in South-east Asia. While AI adoption rates across the region have risen to 14 per cent in 2018 from just 8 per cent the year before, organisations are still struggling to make AI part of their strategic agenda.

According to a study by Cognizant, roughly two-thirds of executives in Europe and the US are enthusiastic about the potential benefits of AI and see AI as extremely or very important to their companies. However, 40 per cent of the same respondents also said that securing senior management commitment and adequate budget was extremely or very challenging. We're seeing similar trends in the Asia-Pacific region. The reality is that despite the widespread excitement, companies are not yet fully committed to giving AI a central role in advancing business objectives.

Business leaders must understand that AI adoption requires strategic preparation, focusing on embedding AI in the organisation's DNA. Data being the fuel for AI, businesses must carefully evaluate data when promoting their AI capabilities, keeping a keen eye on ensuring that technology leverages the right data and capabilities to be truly effective. AI delivers on its promise when large swathes of rich, big data are put to optimal use.

Adopting AI is like nurturing a young student - it needs to be put in the right environment in the hands of capable caretakers, guided along the right path, and taught well when it makes mistakes. It learns as it grows up and requires steady attention for quick progression. To make sure that AI efforts achieve their desired outcomes, business leaders must take its adoption beyond simply building the infrastructure, to instead, transforming from the very core - moulding company culture, updating training programmes, and redefining business goals.

Culture: Spirit of experimentation

Business leaders should develop a culture where experimentation - together with all its wrong turns and missteps - should be welcomed.

The spirit of experimentation embraces the idea that not all initiatives will pan out. Accepting failure is key, but failure in AI isn't binary: Sometimes pilots are technological successes, but yield few benefits.

When Google announced it would shut down its online platform Wave just a year after its launch, it rewarded the development team for having taken a calculated risk. Accounting software company Intuit holds "failure parties" and the company's co-founder, Scott Cook, says that every failure teaches something important that can be the seed for the next great idea.

How can organisations better prepare for rapid experimentation in AI? Ready access to new technologies is an important first step. Many AI efforts get bogged down in lengthy technology procurement times and organisations develop a mind-set that will support experimentation. For example, DBS Bank in Singapore has been promoting the culture of experimentation through various initiatives, including its DBS Academy, that encourage employees to embrace a digital mindset through experiential learning and experimentation.

Additionally, businesses need to be sure they have an open cloud environment to experiment with machine data. Better still, they should create a robust set of partnerships that provide access to continuously advancing AI technologies.

Because no organisation can do it alone when it comes to AI, businesses should examine the good work happening in startups and with other large providers, and determine how collaboration can strengthen their adoption efforts.

Training: Everyone is AI-literate

AI literacy is another significant matter that has to be at the forefront of the AI adoption strategy. In a recent survey on AI enterprise adoption in the Asia-Pacific region, the "lack of skills and knowledge" was one of the most frequently named barriers to AI adoption by executives and IT line-of-business heads.

AI systems require a combination of tools and techniques applied to the right business problems and processes, to deliver personalised experiences with efficiency and scale. Organisations need a certain standard of AI literacy to pilot, execute, sustain, and advance these systems.

The relevant skills and knowledge need to be brought into the front, middle and back offices, as well as higher management. By establishing a baseline of AI literacy, companies can ensure a solid base of employees capable of taking on the convoluted process of AI adoption, and empower the higher management to make informed business decisions and create budgeting allowances.

Organisations could consider crafting appropriate staff training programmes, developing intelligent workflows or subject-specific AI boot camps. They can also leverage government-backed initiatives, such as the Company Led Training Programme (CLT), recently announced as part of Singapore Budget 2019, to deepen employees' competencies for new tech demands for their current and future roles. CLT programmes include training in areas such as cloud computing, data analytics and cybersecurity.

Business Goals: Rethink ROIs

Business leaders foresee significant benefits from AI technology in terms of lower costs, increased revenues, and the ability to introduce new products or services. While these are important and certainly beneficial business goals, there is a plethora of other potential benefits to consider. If done right, AI can help improve product and service quality, reduce cycle time, create new and better employee experiences, and enhance safety, among other things.

Yet, defining ROIs in the light of AI integration is very often a tug of war: The IT department may prioritise AI initiatives with leading-edge technologies that demonstrate proficiency, while business leaders may prioritise achieving instant cost-cutting and time-saving benefits. Companies must find a balance between the two.

The goals and ROIs set should bring a clear focus on finding the right intelligent technologies to solve tough business challenges and ultimately, deliver a competitive edge to the organisation. AI will look different to every company.

For example, nearly half of the financial services companies in the Asia-Pacific are experiencing increased fraud and are looking into AI as a tool that could help them address this challenge. However, many are citing the lack of an overall data management strategy, and unstructured data as key impediments to successful AI implementation. An AI-driven machine learning solution that could enhance fraud detection will enable greater speed and accuracy, while reducing business losses from fraud. The essential thing to remember is that while some AI projects might not result in the explicit answers one is seeking, such projects will deepen the AI expertise and understanding of the larger business problem.

Despite its promise of real business benefits, AI adoption levels in the Asia-Pacific region are still relatively low.

Businesses should realise that when it comes to digital pursuits, there's nothing like AI. Rather than applying their learnings from other digital initiatives, they need to get ready for a whole new way of thinking for AI to deliver optimal success. From rethinking old ways of work, to recognising new types of value, AI requires a fresh look for sure.

The writer is global head, Digital Business, Cognizant.