Implementing AI at scale in an organisation can yield a wealth of benefits, from improving profit margins to saving workers valuable time. But getting vaue out of AI projects requires long-term planning, culture shifts and organisation-wide training. However effective your AI product is at producing insights, it will not return value unless those insights are trusted by the end-users, and acted upon effectively. So how do you lay a strong foundation on which to build momentum and enthusiasm for taking AI to the next level? In their article, 'Building the AI-Powered Organization', for Harvard Business Review, Tim Fountaine, Brian McCarthy and Tamim Saleh discuss the traits exhibited by businesses who have found success scaling up AI.(1)
The authors describe three important shifts that enable organisations to deploy AI at scale. Firstly, employing diverse, multidisciplinary development teams ensures that the AI projects in development meet the needs of the end-user. Involving operational and management staff in the development process will also ensure that AI products address broad priorities, and that potential barriers to adoption are identified early. Secondly, to enable actionable outputs from your AI, there needs to be a shift from top-down decision making to data-driven decision making. The end-user of the product should feel empowered to act upon the insights and recommendations of the product in order to properly employ it. Finally, organisations need to adopt a 'test-and-learn', agile mentality for AI development and deployment. In this model, all end-users become accustomed to trialling AI during its development, and come to see 'problems' merely as areas to be improved in the next version.
These are not insignificant shifts to introduce to an organisation, and the authors encourage leaders to pave the way to successful AI deployment. It is up to leaders to champion the benefits of AI to the workers who may fear AI replacing them, and reassure their staff that successful AI projects will "enhance, rather than diminish their roles"(1). To ensure that the end users of AI products are fully equipped to extract value from them, the authors suggest that 50% of investment in analytics should be spent on "activities that [drive] adoption". This includes training, whether on-the-job or in dedicated 'AI Academies'.
There is no out-of-the-box solution for effecting these organisational shifts, so identifying the potential barriers to change unique to your own organisation is essential. Similarly, deciding upon an organisational structure in which to deploy analytics capabilities and resources should be driven by the individual needs of your organisation, not necessarily by what seems to have worked for others.
By focusing on building a bridge between developers of AI, management and end-users, you can prepare your organisation to scale up AI and reap maximal benefit.
Read the full article here .
(1) Fountaine, T., McCarthy, B., and Saleh, T. (2019) Building the AI-Powered Organization, Harvard Business Review Magazine (July-August 2019 issue), available at http://links.nightingalehq.ai/build_ai_powered_org [Accessed 18 October 2019]