Skills to Take Your Data to the Next LevelThink GradientBlockedUnblockFollowFollowingMar 17Note: I use ‘AI’ as an umbrella term to encapsulate Artificial Intelligence, Data Science, Machine Learning, Deep Learning, and all its other variants.
I know each has its one unique flavors but I’m putting them all in one stew for now.
Big Data World conference — Panel discussionAt the last Big Data World conference that took place on the 12/13th of March at the Excel exhibition center in London, I had the privilege to be invited by HarveyNash.
com to participate as part of a panel discussion and talk about the skills needed to take your data to the next level.
It was interesting to hear the different perspectives from the other two panelists, one coming from Telefonica and the other from Cineworld.
Before we talk about skills, I would like to briefly talk about organizations readiness to successfully deliver Data Science projects.
I strongly believe that without defining the demand first, we cannot talk about the supply.
We are constantly bombarded by the needs of the 21st-century ‘AI-driven’ organizations who always like to emphasize the need for more and more Data Scientists but fail to adequately articulate the reason behind the need for such resources.
Hence, I believe it is important to firstly demystify the cause before we talk about the effect.
What does it take for organizations to successfully fulfill the promise of AI?Every organization is embarking on AI projects and initiatives and it is very important to make sure that you’re meeting business drivers and objectives with any such initiatives.
This is crucial in order to get the buy-in from your internal executive sponsors and keep the momentum of said initiatives going.
It is also crucial from a delivery perspective as with a clearly defined objective and a set of success criteria you can execute and deliver something tangible and that can be of use to the business.
In order to maintain strong sponsorship and support from internal decision makers, you need to make sure that you’re generating value through quick tactical wins.
These could be small projects that tackle a specific use case and that leverage of the shelf solutions from different technology vendors or internally as much as possible.
These tactical projects could be part of a wider strategic vision to implement AI across different business functions (e.
Sales, Marketing, Finance, HR).
However, there are people/process/and technology challenges when undertaking such strategic initiatives.
These could take the form of lack of skills internally, lack of infrastructure or scale of existing infrastructure, lack of good data quality, lack of policies to ensure data privacy and responsible AI practices, agility, team diversity, and many more.
Challenges aside, in essence, these projects could unfold by adopting a three-step approach:Leverage existing technology within your organizations to implement specific use cases within a business function (Complexity: Low)Onboard new off the shelf technology from 3rd party vendors (e.
Microsoft, Amazon, Google, etc).
(Complexity: Medium)Develop bespoke / custom AI solutions (Complexity: High)To smoothly transition to the much-anticipated topic of Data Science skills, from that perspective, I believe the story gets more complicated when an organization has exhausted the first two avenues and becomes reliant on developing AI solutions from the ground up (i.
the 3rd scenario in our list above).
In this scenario, organizations will either look internally for skills or hire an external partner or contractor to help deliver a custom AI solution.
The real question is: Do you nurture Data Science talent within your organization or hire Data Science expertise externally?There are certainly benefits to both avenues.
In the age of AI, domain expertise will increasingly become more important.
I think combining technical know-how with domain expertise, whether that’s knowing the ins and outs of a compliance management system, or preparing statutory filings, or onboarding new customers as part of your HR process, will be a unique attribute for individuals to possess and set themselves apart in the Data Science market.
Of course, automation is a threat to many of these business functions.
Here’s a bold statement that I hope no Super-Intelligent AI of the future finds offensive if it ever reads it: any job that involves the human prefrontal cortex will be difficult for automation to replace it.
This then becomes an internal struggle for business stakeholders who want to infuse AI capabilities within their business function.
A struggle between either nurturing Data Science skills internally or hiring externally.
In the first scenario, it will mean they can leverage internal resources who already know the business and the outcomes they want to achieve in alignment with the business drivers.
Although, this means that organizations will need to identify employees with an inclination towards Data Science and equip them with the relevant technical skills.
This is often time-consuming and costly.
One can argue that teaching a Data Scientist the business and use case at hand is often more economically feasible and takes less time than teaching an internal resource who already understands the business, Data Science.
It is important to find that balance so that you can nurture sufficient Data Science skills internally in order to maintain and operate an AI solution going forward and not become locked in with a vendor or delivery partner, but also hire or collaborate with a delivery partner or contractor to implement an AI solution.
In the panel discussion, we did not go to such details on this topic however these are some thoughts I wanted to share that have inspired me from speaking to various customers out there implementing AI and in general from talking to people from the Data Science community.
During the panel discussion, we also briefly touched on data breaches, data privacy and ownership, STEM, and many more.
I want to leave you with a thought-provoking question that I received from one of the audience members that I found particularly interesting:What role do you see disciplines outside of STEM playing in shaping the development of AI / Data Science?.