Implementing AI with the team you have - DataRobot
Article by DataRobot Asia Pacific GM Tim Young
When companies decide to become AI-driven organisations, the most common question asked at DataRobot is, “Can I implement AI with my current team?”
It’s reasonable for company leaders to assume that embedding AI across their businesses requires specialist skills only held by very qualified - and often expensive - data scientists.Create a Citizen Data Scientist workforce
Data scientists have created most predictive analytics models currently in production, but it’s become increasingly obvious that business users will soon outpace data scientists in this regard.
Organisations need to ensure that AI becomes an integral part of every business process.
Achieving this objective means, however, that employees across the entire business need to adopt AI, rather than expecting it to remain the domain of the IT techies.
The first argument that may be voiced is that most business users don’t have the time or resources to develop the traditional data science skills of coding, deep domain expertise, and knowledge of algorithms.
Enter automated machine learning to bridge the skills gap.Automated machine learning empowers citizen data scientists
Automated machine learning is the ideal tool for building AI and tech skills throughout companies. Global research house Gartner believes that automating data science tasks will not only help data scientists increase their productivity but – in our view, more importantly - will enable analytic-minded business users to cross the skills gap to become citizen data scientists.
Indeed, Gartner sees such a vast backlog of analytical applications, automation offers the only feasible way to address it.
Gartner refers to this trend as augmented analytics.
In truth, automation is just an evolutionary step that all technologies go through.
Teaching the machines to do the work of a data scientist is natural (and expected) although not all data scientists agree (which is also natural and expected).
We’ve held this view about the evolution of the data science market since we launched the world’s first automated machine learning product in 2015.
Recently, other vendors -- Microsoft and Google -- have joined the party and so endorsed the big idea we had over five years ago.
Those businesses that are already data-driven and have teams that use analytics tools are primed for the next step to becoming AI-driven.
Automated machine learning will help close the gap between business users doing self-service analytics and data science.
We’ve seen that everyone benefits from automated approaches.
Those who derive the greatest benefit have deep domain knowledge in sales, marketing, finance, HR and so on, but typically lack the coding skills and detailed understanding to differentiate between each model type.Complementing machine learning with training
Machine learning won’t address your team’s every need as you implement AI.
Like in any job, learning supports career progression.
Your citizen data scientists will need to learn how to build predictive analytics models.
They’ll also need to master AI storytelling to equip them to accurately define AI projects, understand what data to use, avoid bias, interpret results, and communicate effectively.
We’ve helped organisations around the world develop their internal AI competency, including identifying the required skills and recommended citizen data science training for upskilling existing talent.Encourage everyone to identify AI opportunities
An important aspect of integrating an AI-driven approach across your organisation is to encourage all your employees to identify the challenges in the business that could be tackled by AI.
This ensures that your teams understand that they have a role to play, regardless of the nature of their operating role.
This approach may also help you surface hidden issues that may have been ignored or buried when no solution was previously available.
Companies can tap into the deep domain expertise and knowledge of their employees to ferret out these costly blockers.
This engenders buy-in to the process and the knowledge that AI can support and enhance their roles, rather than replace them.