While AI’s impact on the workforce is clearly top-of-mind for most organizations, executives have many additional questions which need to be addressed to clarify the impact of the technology on their business models, processes, products and services – and their market and competitive position under different AI adoption scenarios.
The exact question set varies by executive role. While the CEO is mostly interested in implications on corporate growth, competitive position and differentiation, the CFO may be interested in how much it will reshape the workforce, and the CIO may be interested in which existing business applications and processes may be impacted the most – as well as when and where to make strategic AI investments.
To address these questions, you need to look at how different categories of AI will impact different business functions within specific industry segments. Taking healthcare as an example, if we look at the hospital environment as a sub-segment of the healthcare industry, we can explore how AI will impact the various job functions and tasks conducted in the hospital, the timing and extent of this impact, and the resulting impact on productivity and budgets for specific organizations.
A strategic framework for AI adoption
Before analyzing these various scenarios to understand the pros and cons of being a pioneer, early adopter or early majority player in terms of implementing AI in your specific industry, it’s useful to establish a framework to model the environment and the internal and external forces which can accelerate or delay your adoption.
While your organization is likely conducting pilots and MVPs, and you may have already deployed AI in production in several business areas, the following steps may be useful towards your overall strategic planning:
1. Define your AI taxonomy and AI categories
Firstly, you need an AI taxonomy from which to develop a set of simplified categories of AI which will impact the various business functions conducted in your industry segment. A set of industry job codes will help in exploring the intersection of these AI categories with units of work – i.e. tasks – conducted in the hospital, for example, or other industry segment, resulting in specific AI use cases.
The benefit of having an AI taxonomy is that rather than trying to boil the ocean and determine your AI adoption strategy in a single broad-brush approach, you can segment AI technologies into specific categories – such as RPA, physical robotics, machine learning and more – so that you can craft a specific adoption plan and timeline for each. You can also develop an AI task matrix to understand the impact of each AI category on each business function or task.
2. Establish baseline adoption curves for each AI category
A baseline adoption curve for specific AI categories in specific industries can help with understanding the current state of adoption and the timeframe over which this adoption will progress from the pioneers and early adopters, to the early majority, and eventually to the late majority and laggards. This baseline adoption curve can then be adjusted with a knowledge of the external and internal forces which are either accelerating or decelerating the general adoption timeline.
As an example, one of the AI use cases within the hospital is the use of autonomous delivery robots to serve a wide variety of departments including pharmacy, lab, nursing, food services, linen and environmental services. Robots such as the TUG robots from Aethon already make over 5 million deliveries annually. When compared to manual deliveries, they can help to reduce the cost per delivery (CPD) by 50 to 80 percent.
If we assume the current stage of adoption is pioneer (defined as below 2.5 percent) with 0.5 percent to 1 percent of hospitals already using autonomous robots for deliveries (note that Aethon cites 450 robots deployed to date), we can use this to establish our baseline adoption curve for this specific AI category and industry.
3. Apply corrections to the adoption curves based on internal and external forces
The next step is to adjust these baseline adoption curves by considering the internal and external forces which can accelerate or delay adoption in your industry. Internal forces will be considerations related to your own organization such as your company strategy, culture, financials, investment appetite and digital transformation maturity, and external forces will be government, policy, society and industry factors as well as the levels of investment and innovation within the AI ecosystem.
Data for the internal forces can derived primarily from personal interviews and surveys within your organization and data for the external forces can be derived primarily from market analysis using data from sources such as Crunchbase, Google (e.g. Patents and Trends), Gartner, IDC, McKinsey and other sources. Another valuable resource is Oxford Insights whose Government AI Readiness Index draws from multiple data sources including the Global Innovation Index, UN eGovernment Survey, Tuft’s Digital Evolution Index, Open Data Barometer, and the OECD OURdata Index.
By studying each factor within your model (e.g. we have defined a set of 20 factors which can accelerate or delay adoption of AI in terms of both internal and external forces), you’ll be able to determine whether the forces are “bearish,” “neutral” or “bullish” in terms of impacting the rate of adoption and you’ll be well informed to make course corrections in your strategy.
4. Explore the intersection of AI categories with specific business functions
The resulting impact on productivity and budgets can be calculated by looking at each AI use case in the AI task matrix and determining the timing of adoption from the baseline adoption curve modified by the internal and external forces. The automation potential for each AI use case will enable you to calculate the impact on each specific business function or job code.
As an example, if the delivery role at the hospital can be automated up to 80 percent via delivery robots, this enables a number of staff roles to be reallocated and refocused on higher value activities and priorities such as providing the best patient care possible.
In this healthcare example, results can include improved productivity, improved patient experience, improved worker safety and employee satisfaction, as well as improving compliance. Some of the most interesting use cases for AI are when we apply it to augment human work processes, to instrument the human and socialize the machine, so the two can work seamlessly together.
While success in AI often comes from pilots and MVPs and through learning what works and doesn’t work in real-word scenarios with customers, partners and employees, this framework can help as a strategic guide as you think through the many categories of AI technology that may benefit your business and how and when to apply each within your organization.
This article was originally published via my “Managing Innovation & Disruptive Technology” column on CIO.com.