Easing AI Access: ‘Citizen Data Scientists’ and Humanized Machine Learning 

IDC estimates worldwide data volume is set to rise by 61 percent between 2018 and 2025 – eventually reaching 175 zettabytes – with much of this generated by businesses. So how can this be harnessed to optimize business processes, improve day-to-day operations and inform decision-making? The answer lies with “humanized machine learning platforms,” which are making advanced ML capabilities accessible to business problem owners, enabling the rise of the “citizen data scientist.”

Too Much Data, Too Little Time

Many businesses today are struggling to analyze and extract full value from the wealth of data being generated and gathered daily. The challenge for business problem owners – be they a C-level executive, analyst or even operations manager – is effectively understanding their data to drive further business value and optimize processes.

They may have spreadsheets full of data and use simple data models to extract limited value, but how can they take this further? The answer lies with greater accessibility of ML through user-centric platforms. For the first time, this enables business problem owners – those with intimate knowledge of specific problems and their impact on operations – to connect ML capabilities to business value.

ML has traditionally been viewed as requiring extensive resources, time and technical expertise, which often includes hiring data scientists – a highly specialized field where talent demand outstrips supply. Also, they are often too separated from a business problem to contextualize it and understand the full impact it has on operations.

Enter the citizen data scientists – employees not operating in dedicated data science or analytics roles, who can use a humanized ML platform to explore their data and deploy models to unlock its value. Thanks to user-centric platforms, current employees can enjoy access to ML without the need for specialist training. This is a milestone in empowering data owners to quickly master their own data and complete operations at scale, without significant investment or expertise.

At the company level, this puts advanced ML solutions into the hands of small and mid-sized organizations and their employees. And increased ML accessibility also generates fresh opportunities for data scientists, freeing up their time to get closer to business problems and focus their skill set on innovation for digital transformation projects.

New Business Capabilities at Speed and Scale

A humanized ML platform provides citizen data scientists with greater accessibility to the capabilities required to quickly prepare and visualize data, and subsequently build, deploy and manage a suitable model. Whether this involves suggesting actions to clean and correctly format data or recommending the most suitable model for a data set, a humanized platform is designed to guide users through the process from start to finish.

This approach reduces mundane data preparation tasks. Think of repetitive business processes that involve analyzing data in a similar way on a routine basis, such as budget forecasting. Instead of tying up senior management resources for weeks to finalize budgets, managers can use an intuitive ML platform to quickly identify and set up a model capable of revising budgets annually – dramatically cutting the time investment in this process going forward.

Alternatively, take a manufacturing company that produces precision components. They may have machinery experts with decades of industry experience and deep understanding of the data produced by equipment sensors – but they can’t identify patterns and areas for optimization without a data science team. With humanized ML, these experts can rapidly input, cleanse and visualize data, then select an appropriate data model to uncover previously unseen insights.

Man and Machine: Complementary Capabilities

ML applications are excellent for risk assessment and management, and making data-driven judgement calls, but lack the intuition and creativity required to contextualize and problem-solve for human affairs. This is where humanized ML platforms draw the line between “human” and “computer” tasks. They take on the labor-intensive, repetitive work, such as data cleaning, data-driven model discovery and model validation, and empower problem owners to focus their time and resources more directly on the business problem at hand.

Ultimately, the computer will always have to collaborate with a human when applying ML. To ensure project success, ML needs to form part of a human team, augmenting human skills, intelligence and capabilities. Humans have the unique capability to contextualize data and associated errors. Take a simple example where error codes are present in a large data set. An ML platform will struggle to contextualize this, but a human who is close to the business process can quickly provide an explanation, such as sensors being out of range.

Beyond the immediate benefits, ML platforms solve the issue of legacy when a citizen data scientist leaves the company. These employees can develop ML solutions to solve specific business problems, secure in the knowledge these accomplishments will still be operational, intuitive and reusable by colleagues once they have moved on.

Viable ML for Every Business

ML is set to become increasingly common among businesses of all sizes as they push to optimize daily operations. Don’t forget, business problem owners will always have a unique and intimate knowledge of a specific problem and its relevance to existing business priorities. For the first time, they can directly identify and enhance the value of their data by quickly harnessing machine intelligence at scale.

Nathan Korda is director of research at University of Oxford machine learning spin-out Mind Foundry.