We know what it’s like to introduce cognitive computing to a large organization. You’re 12 years old again, walking into an amusement park. It’s overwhelming: Thrill rides to the left, screaming kids to the right and bright, shiny distractions as far as the eye can see.
Where do you go first? How do you avoid expensive wrong turns? We advise starting small, using agile development to create quick proof-of-concept projects that reveal the potential of artificial intelligence and machine learning (AI/ML).
Why not just go all-in and transform the entire enterprise? Isn’t technology evolving so quickly that there’s no time to lose? These are legitimate concerns, but you have to weigh them against the immense complexity of AI/ML projects. The road to AI/ML has five phases:
- Data integration. AI/ML requires multiple massive databases. You have to ensure your data is accurate and well organized. Missteps here corrupt everything that follows.
- Reporting. Dashboards help everybody in your organization visualize the opportunities to drive value from data analysis.
- Data exploration. Data scientists dive deep into the insights from all these information sources, looking for the best opportunities to create predictive outcomes.
- Predictive modeling. Algorithms use statistical models that seek out successful results and reject failures. Repeating this process through automation allows the algorithms to essentially teach themselves to predict future behaviors.
- Prescriptive modeling. Organizations gain insights into the AI/ML models’ behaviors and use them to guide future operations.
Let’s say you’re an online retailer with a call center that generates one-fifth of your revenue. An AI/ML proof-of-concept project could analyze call-center conversations to predict fluctuations in consumer demand and optimize inventory control. Working the bugs out here provides a model for future AI/ML projects.
Why You Need Human-Centered Design in AI/ML initiatives
User-centered system design starts with the technology touch-points people need to get things done. It means looking at mobile devices, web apps and other touch-points, analyzing how people use them and finding ways to improve their experience.
The user provides an essential point of entry on AI/ML initiatives because people and their devices generate oceans of data you can parse in myriad ways to anticipate future behavior. Conventional statistical modeling can predict when a component in a factory production line is about to fail. By contrast, an AI/ML initiative could conceivably scan the behavior of millions of consumers, predict their response to your new product line and help provision logistics for the next six months to handle the increased demand.
Human-centered design is one of the pillars of DMI because it flows from the innate desires of the people who use technology. When you put users first, they reward you with better engagement and adoption, which creates a virtuous development cycle.
We also use agile methodologies to get prototypes up and running quickly. User and client feedback accelerates the testing and iterations of a development project.
DMI possesses a wealth of deep experience in helping large organizations implement advanced analytics, big data and cloud integrations quickly and efficiently. Our data-science team is bolstered by engineers with wide-ranging expertise in system design and implementation.
Moreover, we’ve worked with global corporations, retail chains, industrial companies and government agencies. We also provide strategic guidance at every phase of a project.
These are the prerequisites for success in an AI/ML initiative.
It’s imperative to understand artificial intelligence and machine learning are not shortcuts to digital transformation. You have to carefully choose the best place to introduce AI/ML, and you need a strategic approach to expanding its benefits across the enterprise.
Ultimately, AI/ML cannot coordinate the people, processes and technology platforms you need to transform your organization. That’s what humans do. And human-centered design ensures that it gets done right.
— Venkat Swaminathan, director of AI and IoT