Only one third of artificial intelligence (AI) projects have succeeded. There are 3 key reasons that explain the failure of artificial intelligence projects:
1. Not appropriate time spent on strategy creation before jumping into execution.
2. Wrong process selection – AI works with large data sets and requires a lot of process experts, most of them do not exist in many project teams. The project becomes a data creation or cleaning project rather than AI Implementation.
3. Badly defined team & roles – Most of the companies just focus on data scientist roles and hire experts with Python & R skills, while in the whole life cycle of AI project data scientist’s contribution is less than 25%. While rest 75% are not integrated into the team and all work in silos.
We should follow the 90-90-90 principle for AI projects. The premise behind this principle is as follow:
1. Greater than 90% of machine learning opportunities are for supervised learning, which revolves around classification and regression.
2. Greater than 90% of work involved for supervised machine learning in streamlining the process, data collection, data cleaning and identifying variables.
3. 90% of automation projects can be done without artificial intelligence tools and costly infrastructure. Simply use your six sigma teams and advanced automation techniques like robotic process automation (RPA).
Most of the businesses already have strong process standardisation, Data analytics and RPA teams and does not have huge data sets like telecom, e-commerce and healthcare companies. They should first exhaust their efforts with above investments to maximise benefits, especially when now RPA tools are integrating AI capabilities with cloud based solutions. They are also more cost effective than full fledged implementations.
Leave your comments
Post comment as a guest