Businesses are quickly realizing the potential for achieving competitive differentiation by introducing artificial intelligence (AI) technologies into their organizations. Externally, they can achieve this competitive differentiation via products and service enhancements. Internally, they can achieve this differentiation via streamlined and efficient business operations. In the end, it is all about time to value of insights creating top-line and bottom-line impact on the business. Not acting quickly and sufficiently on AI investments is a risk that no business can afford to take.
Investing in AI is quite different from procuring and deploying other enterprise off-the-shelf products or services. In fact, it is a highly bespoke endeavor that requires data scientists to work in unison with business analysts, data engineers, DevOps and IT, and app developers to define and implement an AI strategy and launch AI initiatives. Data scientists can develop AI models that map to specific business problems or imperatives; developers with expertise in building custom and cloud-native apps incorporate those models in a production application. So far so good.
Operationalizing AI is another story altogether. Two recent IDC studies provide some insight. One found that more than 30% of the respondents cited a failure rate of two-thirds for their AI projects (source: IDC’s AI StrategiesView, April 2021). Another found that 80% of the respondents cited an average duration of three months to one year spent on building an AI model for deployment. Further, these respondents spend up to a year preparing a completed model for deployment (source: IDC’s AI InfrastructureView, August 2021). The reason for these outcomes is that most businesses do not develop AI apps in the same manner as conventional enterprise apps. That can present a problem during production rollout.