What businesses should consider when implementing AI
- 07 October, 2020 14:24
With the NSW Government releasing its first-ever Artificial Intelligence (AI) Strategy and COVID-19 dramatically increasing the pace of digital transformation and the need for AI to enable that change, many businesses are wondering how to navigate the rapidly evolving technology landscape. Moreover, with AI a clear C-suite imperative, businesses must apply AI solutions in smart, economical, and ethical ways. Most importantly, they need to be human-centric.
With each new application of the technology come new and unique challenges, so measuring AI’s success is key to proving return on investment (ROI). In light of this, here are five things businesses should consider when implementing – or looking to implement – AI into their operations.
1. Focus on the wins
When starting out with AI, organisations often make the mistake of tackling business challenges that generate the greatest return or make the most dramatic business impact. However, speed is just as important as size, and there are always smaller tasks that will generate value quicker, such as automating manual data entry or predictive maintenance. Achieving nimble value fast is the key to effective digital transformation.
One thing to keep in mind is that ROI must always play a role in decision-making. Never should the cost of a solution exceed the value of its benefits, so forecasting the expected ROI of any solution, big or small, is important.
2. Failure is the best teacher (but keep it small)
Another reason businesses should start small is that it is rare to get AI right the first time. It can take several iterations and continuous course corrections to achieve a satisfactory level of accuracy, which means setbacks need to be corrected quickly. Without mistakes, there is no improvement. Therefore, you move faster towards success when you focus on smaller problems first.
It is important to continually assess whether a solution remains valuable and viable. Frequent iterations are desirable when implementing AI, so don’t be afraid to drop an idea and start over. Once a small problem is solved, and has a positive impact on the business’s bottom line, the solution can be scaled up to solve bigger problems.
3. Involve key stakeholders in the whole exercise
Agile, dynamic squads that involve stakeholders throughout the lifecycle of an AI project make for the best approach to success. Bringing the entire scope of enterprise capability to decisions on priorities and trade-offs is the cultural mix best suited for AI success.
Understanding the expectations of an AI-driven transformation and aligning the exercise to address those needs are vital to effective change. The art of transformation is driven by engagement, alignment and agility, so involving the entire organisation from the very beginning and maintaining a dynamic program is the most important operating principle.
4. Data is AI’s most valuable resource ― cultivate it
Businesses have been building data for months, years, or even decades. Harnessing that information is key to AI. Supplementing that data with more and varied sources further enriches the insights available to an organisation. Cultivating and curating a rich and varied pool of information will prove an invaluable asset, which will unlock numerous opportunities to experiment, improve and transform business.
In our experience working with a wide range of organisations, we have found that businesses that maintain modern data platforms achieve business value 40% faster than legacy approaches to data cultivation. Businesses are able to increase revenue, remove business silos, and increase service levels, while simultaneously decreasing the costs of storing and managing enterprise data.
5. Define what success looks like
IT and AI projects are often placed in the same basket, but they’re fundamentally different. AI projects are an exercise to understand the unknown, often exploring opportunities where it is impossible to know the specific outcomes ahead of time. This calls for a different approach to measuring success.
By design, AI projects achieve the best outcomes when they undergo several iterations. Therefore, when measuring success, businesses should focus on outcomes with a focus on AI’s impact on specific tasks and how much value that generates for the business.
It is never wise to implement AI to merely “tick a box”. AI is an investment that can generate profound returns, but only if it is purposeful and well-aligned to business need. AI can unlock countless possibilities and it can be easy to stretch your enterprise thin, so focusing on what is valuable and achievable is always the best place to start. It is also important to remember that a fear of failure is a fear of learning, and AI projects ultimately cannot progress without the lessons learned from mistakes.