4 Critical Questions to Ask Before Starting an AI Project
More businesses are taking on AI projects, but many still aren’t finding success. Here’s what you need to know before taking on your first artificial intelligence project.
If it feels like everyone is implementing artificial intelligence, it’s largely because they are. AI projects are poised to double this year with 40% of companies deploying AI by the end of 2020 according to Gartner. Statistics like these can create pressure on CIOs as their executive team wonders why we aren’t innovating in this space.
Underneath all the optimism and hype around artificial intelligence lies a harsh truth. A study by MIT-Sloan/BCG found 65% of companies reported seeing no value from their AI projects. With value being elusive for so many, how can we beat the odds to deliver success for our business? Let’s look at four critical questions you need to ask before taking on your first artificial intelligence project
People hear every day how artificial intelligence is revolutionizing business. While that’s true, starting a revolution shouldn’t be the goal of your first AI project. Instead, target a small project that can deliver a quick win. Success breeds confidence and can set you on a path for continued success.
With that first project, you are looking to cut your teeth by gaining knowledge and showing how AI can make an impact on your business. Choose a project with visibility at the highest levels of the organization. Find something that closely aligns with existing business processes so that impact can be felt. When you deliver the project successfully, shout it from the rooftops, and find ways to reward every contributor who made it a success. You want AI to become infectious throughout your organization where department heads start asking how this technology can facilitate impactful change for us.
2. What does your data look like?
AI and machine learning hinge on data — lots of it. We need to analyze our data store to see what limitations might hinder our project. Is our data skinny? Is it dirty? If it takes years to adequately compile enough data, the project isn’t viable. If our data is a mess, we have to determine what effort is involved by our data scientists to cleanse it.
Regardless, perfect data doesn’t exist, and we can’t let that hold us back. Don’t settle on a low-impact project because another dataset is more complete. The discovery stage is the perfect time to jump in and explore what you have. Take some time to model the data to determine if you can tell the story with less.
