This post was previously published on LinkedIn Pulse.
Machine learning becomes the NEW NORMAL as more and more companies are embarking in such projects; therefore, it's important to distinguish what's hype (i.e. gaps) and what's real value for businesses.
Below you can find five such gaps which will allow you to plan and execute your machine learning journey in a safe manner, with valuable outcomes for your business.
The reality is that companies don't have machine learning problems. There are just business problems that companies might solve using machine learning. Thus, it is essential to identify and articulate the business problem before investing significant effort in the process (i.e. time, budget, people). A good practice is to prioritize the cases to be addressed based on the estimated business benefits.
In practice, there is no magic in machine learning, and the path to ML success is hard - it takes time and effort, especially at the enterprise level. Machine learning needs a process and a powerful framework to be successful. Data and models are just some of the ingredients, but enterprise ML needs a holistic and uniform approach. In order to industrialize ML, companies should address a new challenge - ML Ops.
Data and algorithms quality and relevance to the defined business problem matter most. The quality of the inputs (in terms of completeness, accuracy, context) determines the quality of the ML outcome applied to a business challenge. In addition to quality data and relevant algorithms, the other two key elements for success are usage of the latest technologies (hardware and software) and the right mix of specialists (data scientists, developers and business owners).
In practice, a combination of top-down and bottom-up strategies will ensure the right balance. On one hand, it's important to create the ML vision and strategy, to ensure sponsorship at top level. On the other hand, it's equally important to have the right mix of specialists who will build prototype projects in different lines of businesses. Thus, learn by doing in the context of your organization to ensure a strong path towards success.
The fact is enterprise ML and people need each other. The new paradigm of machine learning is considered to have a similar impact for humanity as other major discoveries had in the past (i.e. electricity or engine discovery); thus, a similar trend is expected for human labor in the ML era. Some of the jobs will be reinvented, but for sure, machine learning will augment people. Repetitive and unpleasant tasks will be covered by ML applications, while people will focus on value-added ones and thus get additional professional satisfaction.
In conclusion, the gap between the ML hype (the promise) and real value is significant. The imminent risk is that top management might set the expectations at hype level, but the team needs to execute ML at the maximum real value level; this may lead to a lot of dissatisfaction from both sides. To bridge the gap between both sides, you should strike a balance between desired business outcomes and ML challenges your team may face.
Thus, this Latin proverb fits well in this context: "Festina lente," which means "Make haste slowly." In other words, if you can Mind the Gap between what's hype and what's real value, you'll unlock valuable insights to drive better business outcomes.