Hardly a week goes by without a story somewhere about a new start-up or firm that’s disrupting their industry through the use of AI, or an analyst report or industry expert bemoaning failed efforts and missed ROI. Maybe that’s because both things do happen. With AI getting so much attention, it’s not difficult to find real life examples of when AI projects succeed, or when they fail. The outcome of which is largely driven by how firms approach AI-driven innovation. Which leads us to the important question, how can you make sure your AI project ends up on the plus side of that ledger when you’re starting out?
We have some tips for you, and a webinar you can view to get more details.
So, let’s get started.
1. What questions do you want to ask?
It all begins with a series of questions. Before you can begin diving into your data and unlocking insight, you must first work out what it is you want to find.
You can start to define this by writing out the top 10 questions you’d like answers to in order to better understand your customers, or empower your employees, or improve product innovation, etc. It’s important to get this first step right. These questions lead you to the data you want analyse, the analysis will provide you with insight and in turn this insight will provide you with actionable intelligence and your next steps.
2. Determine what data can answer your questions
Once you’ve defined the questions you seek to answer, you can set about identifying the best sources of data to use.
Unstructured data sources come in many shapes and sizes, from both within and outside the company. Some of the most common include social media, product reviews, blogging platforms and call or chat transcripts.
Determining which data sources will likely be the most useful to start can save time. But remember, applying AI to solve problems is an iterative process, and you’ll learn valuable things about your company’s data, and what it contains, at every step.
3. Line up help: 1-2-3
Now it’s time to start building your insight engine with Watson Discovery AI, to begin leveraging that unstructured data.
Watson Discovery AI is designed to be accessible, meaning you’ll most likely have the skilled resources you need in the organization to get started right away. You’ll need…
• A developer who understands RESTful APIs
• A subject matter expert who understands the questions you’re answering, the data needed to answer them, and where that data resides
• A data expert who understands where and how to access your data, company security and personal identifiable information (PII) standards as well as the key performance indicators (KPIs) needed to provide perspective
4. Determine where your analyzed data will reside
Watson Discovery is hosted on IBM’s Cloud platform — offering a range of plans to suit your budget, query and data needs.
Wherever your data resides — on-premise databases or alternative cloud platforms — IBM Discovery is capable of calling and querying your data.
Uploading it directly will allow you to make full use of Watson Discovery’s NLP capabilities.
While Watson is owned by IBM, the data you bring is secure, and that data, the models you train and the learning you uncover is yours alone.
5. List your new questions and expand your exploration
Answering the first set of questions is just the beginning on a journey of discovery with your newly built insight engine.
The answers to your first questions will likely lead to new questions and opportunities for deeper insight. Watson Discovery enables “augmented intelligence” and enhances as well as scales human expertise, helping you uncover new insights thought to be impossible. Now is the time to innovate and iterate. Adjusting your current analysis, expanding the scope of analysis or building a user-friendly display are just a few of the starting points on your continued exploration.
Check out the Watson Discovery on-demand webinar for more details.