Can expectations about emerging technologies sometimes be too high? There’s little doubt that artificial intelligence (AI) is going to change our world, with its power to look beyond what humans can assimilate and come up with entirely new linkages and outcomes. The bigger questions are – how and when? Has AI been oversold so far? Since we’re already surrounded by AI and machine learning (hi Siri and hey Alexa, thanks for listening in), when AI finally “arrives”, will we ultimately be disappointed?
High expectations are perfectly understandable. Extraordinary potential outcomes are on offer from AI. For example, we’re already seeing it used to reduce workplace stress, to find cures for diseases faster, and to reduce environmental impact through smart traffic routing.
Business hasn’t been slow to grasp the implications: new research carried out by Forrester Consulting on Fujitsu’s behalf shows 86% of global business decision-makers already recognize the need to master AI to drive business success in the next five years. They’re purposefully driving AI into the workplace to increase operational efficiency and effectiveness, to lift employee productivity, and free people from the sort of monotonous tasks no one actually wants to do. Instead, with technology taking care of the repetitive tasks, people can focus on other, more strategic activities that produce greater value. Along the way, AI is improving customer experiences and satisfaction with faster, 24/7 access to services, information, and support.
Companies are starting to question the value of AI implementations
It’s fair to say a lot of companies are starting to have questions about the true value of AI implementations. They seek greater impact and faster time-to-value. They worry about running endless pilots just to understand the potential and they’re all too aware of the lack of skills and the high costs associated with AI.
They’re also increasingly uncertain about the AI concept itself, with more fundamental, existential questions now top-of-mind. Do we understand how AI reaches its conclusions? Can it be trusted? Was any bias introduced in its training? Or does it pose risks when it comes to legal liabilities, customer relationships, and employee morale?
Given the extraordinary benefits on offer with AI, allowing this creeping sense of disappointment to proliferate risks ‘throwing the baby out with the bathwater’. Fortunately, there are solutions at hand.
How to get even more value out of AI investments
There’s no magic equation that translates AI resource and effort into dollars in the bank. But there are common sense principles that you can follow.
- Focus on outcomes, rather than technology
There’s no ‘one-size-fits-all’ on offer here, but most of the companies we’re engaging with are looking to improve operational efficiency, maximize revenue streams, track and stay ahead with new trends, and develop new business models to better support their operation needs and scale their business operations. Examples we’re working on include the use of AI to reduce fraud during self-scanning of shopping in supermarkets, fuel and emissions reduction for shipping fleets, accurate detection on brain aneurisms, and more precise credit scoring to approve loans, credit analysis, screen and service prospective borrowers efficiently. There are many more I could mention, but these give the flavor of the sheer diversity of potential use cases and point to the absolute necessity for co-creation to bring together business and technological understanding.
- Test whether AI creates value before committing to scale out
It sounds obvious, but it’s surprising how often this doesn’t happen. At the outset, insist on a rapid proof of business (PoB) engagement, as it’s vital to know in advance whether the investment can be justified in terms of return on investment (ROI). These are usually delivered inside two weeks, as opposed to a proof of concept, which can take longer and tends to check only whether technology can perform as anticipated.
- Look at the wider picture
AI doesn’t happen in a silo and it’s not something you add on externally. It must be embedded in making everything you do better – helping all processes become smarter and generate higher value returns. AI is many things and should be woven throughout a company’s business, rather than hived-off and treated as a ‘side project’.
- Lift value is an end-to-end and modular agenda for AI deployments
No single vendor can ever provide every aspect of critical infrastructure and services for an AI deployment. Therefore, ecosystem partnerships should be a fundamental characteristic on any criteria list, alongside the ability to select the modules required, rather than a bloated offering with many redundant features.
There’s a lot of ground still to cover. My next instalment will dive into how to make AI payback faster, how to tackle the skills shortages that pervade AI projects, and how AI explainability is coming into focus. I hope, by the end, to have helped allay any sense of anti-climax you might have had about AI in the enterprise.