Convince the business, in terms they can understand.
Imagine you have been selected as a proponent to bring artificial intelligence into your business. Are you going to talk to executives about algorithms, training data, test data, supervised learning, unsupervised learning, and deep learning neural networks? Watch their eyes glaze over.
Not that business leaders aren’t getting the AI message. There are plenty of AI efforts underway — the global AI adoption rate grew steadily and now is 35%, a four-point increase from the year before, a recent IBM study shows. It’s clear, as shown in the study, that there are tangible benefits — half of organizations are seeing benefits from using AI to automate IT, business or network processes, including cost savings and efficiencies (54%), improvements in IT or network performance (53%), and better experiences for customers (48%).
But bringing in AI means changes for the organization, and the systems and processes that have defined the way business is done. Selling AI as a new way requires some degree of empathy, finesse and business sense.
The following are recommendations provided by industry experts who work day to day to advocate greater AI adoption:
Think and understand the problem to be addressed before bringing in the technology. “Instead of coming in and asking for a blank check to tackle numerous vague goals, identify a single problem and outline a clear strategy around how AI will solve it,” says Sagar Shah, client partner at Fractal Analytics. “Highlight the quick wins that can be achieved in the first 12 weeks of the project. This provides clear parameters for skeptics to compare against and judge results.”
It’s a matter of taking more time to understand what problem needs AI applied to it. “The single biggest inhibitor to AI success is relying on unscientific assumptions and broad goals, versus taking the time to frame the exact problems they are trying to solve,” Shah says. “Simply put, the more time a company spends with a problem ahead of time, the better AI product adoption will be.”
Set realistic expectations. “Set realistic expectations of what AI can accomplish,” says Alex Saric, smart procurement expert and CMO at Ivalua. “It can be a powerful aid, but algorithms still have a long way to go. Setting realistic near-term goals and working to improve over time is important. Leaders must sell AI in a way that explains it as helping simplify work, and enabling employees to focus and become more engaged and productive, rather than displace them.” Often, people assume AI means automatic innovation, versus “really understanding what it means for them and how it will be applied every day,” Shah agrees. “At times, a more traditional approach may be more effective than one powered by AI, so buyers must seriously look into the details.”
Get talent on board, but don’t try to build everything yourself. While AI is seen as a labor saver, the skills gap remains the biggest barrier to AI adoption, another recent IBM study shows. “The talent gap is significant, particularly when considering state-of-the-art algorithm development in AI,” says Flavio Villanustre, senior vice president and global chief information security officer for LexisNexis Risk Solutions. “It is usually not cost effective for a company to develop their own algorithms, unless selling services or products using those algorithms is in their core business strategy.” Too many companies “venture into forming an internal team to build AI solutions end to end,” says Sivakumar Lakshmanan, co-chief executive officer at antuit.ai, now part of Zebra Technologies.
Think long-term. “Another common misconception is that businesses need to adopt AI all at once and that every project needs to be an immediate success, otherwise their investment has been wasted,” says Shah. “Adopting AI is a big step for any organization, so to start, focus on applying AI to a handful of strategic problems – of which only a few will work immediately as anticipated – and then take the lessons you have gathered and apply them to another set of projects, and then another. Businesses also need to keep in mind that AI is not just a single tool. Instead, it is a culmination of ongoing engineering, design and behavioral science work. All of these elements need to work in concert.”
Rethink processes. AI simply cannot be layered on top of existing processes to expect success. A common mistake is “fitting AI on top of an existing process which is centered around manual, committee-based decision making,” says Lakshmanan. “In this case, AI ends up being a tick in the box. Instead, adapt the processes to the newer world – for example, you don’t buy an e-book on a device to print out and read.”
Ensure transparency and trust in data. “How AI output is generated must be made more transparent by the technology itself,” says Saric. “Consider that AI has been used for many years to help classify invoice lines from company purchases in order to accurately determine where budget is spent. But many approaches are a black box and when users inevitably find mistakes among millions of lines classified, they lose trust and stop relying on the data.” The quality of AI output “is highly dependent on data quality and volumes but substantial amounts of organizational data is still dispersed among many systems and is of dubious quality,” he adds. “ Organizations need to gain control of enterprise-wide data to reap the potential of AI. Building a solid data foundation, leveraging master data management solutions and platforms with unified data for specific functions — suppliers, customers — can address this but are not broadly in place today.”
Involve employees at all levels. “Businesses fall into the trap once they have identified the problem of steamrolling a tool into production without significant amounts of collaboration and end-user input,” Shah says. “To drive the adoption of AI tools, businesses need to co-create them with end-users so that they don’t only solve a problem in theory, but also in practice. Even with robust co-creation and problem framing, sustained adoption hinges on businesses and their AI partners constantly working on enhancements. New needs and wrinkles appear all the time. And the only way to make sure that your company is set up for long-term success is to accept that AI is a living breathing tool that needs constant tweaking, not a one-time plug-and-play solution.”
Convince the business, in terms they can understand.