For those of you who did not read the first few articles in this series, you can find them here (part 1, part 2, part 3).
This week, I will take the enterprise point of view to answer the same question that we addressed in the past weeks from a student perspective.
So, this week, our original question, “how do I become a data scientist” changes to
How does my organization adopt data science?
Today, this question perhaps ranks top in the minds of many CXOs across the globe. I get it so often that I made this a research area and currently working with a few doctoral students on this!
I notice, even today, many organizations believe that hiring a strong team of data scientists, giving them adequate funding is all that is needed to adopt data science.
In our experience, the most successful AI companies followed a different path.
Firstly, they accept that Data Science is not just a specialization of computer science. It is a general tool to solve a variety of business problems”.
They believe that such alignment in belief has to happen across the organization, and it is best to start from the very top. They clearly define what is expected of each member and create mechanisms to empower them.
The Top Rung
To adopt data science, they do not need to learn coding in python! They should be empowered to
Decide whether they intend to use AI as a differentiator or an efficiency driving engine
Create short term, medium and long term AI priorities
Implement AI friendly policies
Understand legal, ethical implications and ensure they are adequately addressed
Most importantly, they should understand what kinds of teams must be built across the organizations for successful adoption
The Middle Managers
They need not learn to code in python too!
The most important thing they should internalize is that they are an active part of an AI implementation and not a passive receiver of a solution developed by a tech team.
They have to contribute meaningfully in design, development and implementation. After all, it is their business, their ROI and their teams that get impacted either way because of AI. They must be
Empowered to speak data science language and translate business problems
Understand the importance of the data
Define business aligning performance metrics
Design production plan (people issues and technical issues) for AI implementations
Nowadays, these guys have received a nice title. They are called “Translators”.
IT teams and tech-friendly biz team members
These are not your true math and data geeks. How many of them can you hire anyways?
There are excellent free and commercial tools available nowadays to solve a variety of data science problems without any mathematical or coding background. They sort of work as plug and play.
The tech-friendly guys who can code a bit can be ramped up quickly to use these tools and solve many day-to-day problems using data.
It is important to train a large group of your tech-friendly teams on such tools and on correct and disciplined methods of building data science models. This ensures two things.
You solve many more problems using data science, provide the right insights and
Your data science team can focus on solving really tough problems that truly deserve their attention
The Non-IT executives
These people do not want to become data scientists. In fact, many of them may hate tech stuff actively. However, they need to learn a few things too.
They should be trained to:
Collect data well so that it can become an asset in the organization
Accept predictions and apply them even if the machine does not agree with their intuition
Analyze what tasks in their jobs can be AI-enabled to enhance the efficiency
Many times, in our experience excellent data models go waste simply because people show enormous resistance to use.
There is also that insecurity and discomfort of machines taking over!
Yes. The geeks. They build the new age products and services, do the R&D, stretch the algorithms and all the cool stuff. They need to
Understand how to scale and deploy products such that they actually drive true value
How to effectively read, interpret, challenge and conduct research
Hopefully, it is clear that for an organization to adopt data science, every member has a definite role to play.
Organizations that implement such structural role definitions and construct an active AI environment are more likely to succeed than those with a few brilliant data scientists.
BTW, we spend a lot of time thinking and solving this problem. If you are looking for some organization to help you successfully adopt AI, we will be glad to talk.