Companies and governments have evolved their questions about artificial intelligence from “What is AI?” to “How can we use AI?”
In fact, businesses leaders are now seeking justification to use AI in their operations, IT leaders are asking for budgets to implement AI, and executives are calculating the benefits AI will offer.
And yet there is still hesitation when it comes to implementing AI. Gartner recently surveyed about 80 members of its Research Circle to see where they stand on AI adoption. Most of the participants in this sample said they are still gathering knowledge and developing their strategies.
Certainly, organizations would do well to make sure they are not just chasing what looks “cool” and instead determine how AI can best serve their purposes.
In terms of the tools currently available, AI is particularly effective at answering questions that are commonly asked but whose answers vary depending on context. Examples of this kind of question might be “What should I wear with this shirt?” or “How can I close this deal?”
Look before you leap
As businesses move forward with the process of identifying and implementing useful AI technologies, it’s important that their chief data officers (CDOs) are included in any evaluation of whether a project is suitable.
In the same Gartner survey on AI planning, only two of over 80 respondents said their CDO had initiated their AI efforts or was responsible for technology decisions. A CDO can help prepare a company for AI integration, set project priorities, and generate value broadly throughout the organization.
When evaluating an AI solution, organizations should use conventional requirements that each vendor will be measured against. Metrics should be measurable and realistic. Most AI projects are still evolving and investments should be seen as furthering an organization’s understanding of the technology.
Lay the groundwork
Regardless of the specifics involved, it is important to ensure that any projects AI is applied to are well-suited to an exploration of the technology. Teams should be able to match the following attributes with a potential AI integration project before deciding to move forward with it:
- High-quality, credible data is essential. Organizations should invest in efforts to verify and improve the quality of data for AI integration purposes.
- Data should have clear parameters that are reliably followed. An effective analytical strategy relies on the ability to establish a multivariate view of the data that is being analyzed. It also requires the ability to reduce elements, such as transactions or interactions, to parameterized vectors. Having too few parameters or inconsistency in content will create insurmountable obstacles.
- Demonstrable value to stakeholders. Initial pilot projects should target existing needs.
- Ascertain that the organization’s AI-related goals are reasonable and possible. Service providers, in particular, should be able to provide examples of similar past projects. If an organization is lacking firsthand proof, it should seek out academic research indicating that the goals set in place for the potential AI integration are viable.
All of the organizations Gartner spoke with in the study indicated they have or are planning to integrate AI into their existing customer engagement practices. The three most frequently cited application categories for such integrations were related to customer interactions. Here are some highlights:
- One in three organizations said they will link AI to customer engagement applications.
- Three in 10 said they will integrate AI into call center service and support processes.
- One in four said they will integrate AI into digital marketing operations.
Who stands to gain
There are several aspects of customer relations that could particularly benefit from AI applications. Areas where companies should seriously consider integrating AI include:
- Sales: Advanced analytics technologies work well for developing sales strategies. Companies should consider initially investing in machine learning to automate the personalization of marketing and merchandising messages.
- Customer service: Initial implementation of virtual customer assistance should focus on a small percentage of customer interactions that are easy to automate. Companies should choose use cases with the greatest possibility of automation. Selecting use cases with the lowest value for transactions and/or the most potential for increasing value is also ideal.
- Office and collaboration suites: Companies should use cloud office suites to expose a broad employee base to AI benefits. Some of these benefits include information awareness, task automation, and expertise augmentation. The digital workplace program should be a hub for educating the workforce about AI.
- Supply chain management and manufacturing: We initially use AI to discover gaps and anomalies in communications and processes with suppliers. In order to reap the full benefits of AI, businesses must work toward correction of such problems once they are identified.
- Communications: Gartner predicts that poor-quality implementation of artificial intelligence could significantly decrease customer satisfaction. This is why it is important to identify narrow, routine customer interactions for AI integration. It is also helpful to test the ability of virtual customer assistants to handle these interactions before fully integrating the new technology.
- Government: Any governmental organizations should initially invest in smart advisors for internal and intergovernmental applications. There should be potential to expand applications to include constituents.
- Education: Schools should initially invest in operational improvements, such as student/teacher pairing and improved evaluations of structured work. AI also promises to relieve teachers of some of their most tedious tasks in the future.
- Retail: Retailers that have embraced big data analytics may have become deft at uncovering insights. AI offers the potential to react to such insights in a more agile fashion. This allows retailers to potentially achieve more successful personalization for in-store and digital commerce.
- Banking: Banks are evaluating and deploying AI-automated interactions between customers and banking systems. This is taking place predominantly via digital assistants. Such bots might be voice- or text-driven.
- Health care: No vertical presents greater visibility for AI than health care. Health care providers should initially focus their efforts on automating the practice of identifying errors in labor-intensive processes — especially if such processes are costly.
Whit Andrews is a VP Distinguished Analyst for Gartner, Inc., focusing on artificial intelligence and cognitive computing.