From large corporations to smaller upstarts, AI technology is being introduced and adopted by organizations across industries as it moves from its academic roots to the forefront of business. But before thinking AI, think data first.
Jul 08, 2019
From Amazon Alexa to Spotify and the Nike Fit app, artificial intelligence (AI) technology is becoming a household norm. There’s not a company today that isn’t looking to leverage AI to deliver new digital services and insights that disrupt, or fend off disruption within their market. Every company is a data company, whether they know it or not.
But AI is only as good as the data you feed it, and access to quality data can become an Achilles heel. The only thing worse than bad decisions are good decisions based on bad data. A winning AI strategy requires developing not only a robust platform to manage the end-to-end data lifecycle but changing how teams work with data.
To deliver success, companies must build a culture that puts data front and center. This is more than just making data-driven decisions, but building trust in data and facilitating collaboration between teams. To accomplish this transformation, companies need to invest in technology that provides rapid secure access to quality data wherever it’s needed.
The data lifecycle doesn’t start with your production databases, nor does it end with your data lake. Data is born at the edge - your users, your sensors, your employees. Success starts with understanding what data you're gathering and why, with guard rails to ensure that individual teams are making good decisions that prioritize privacy and compliance at every turn.
The world's biggest tech companies like Amazon, Facebook, and Google have proven their massive trove of data can be leveraged to drive differentiation through AI. But all that data would be for naught if the people who need it can’t get access to it. Solving this requires facilitating the flow of data everywhere it needs to be. Not just to application development teams and analytics teams within the company but also to third-party services outside the company.
All this data access quickly becomes a security nightmare. Not only do you need to intimately understand where sensitive data exists, you now need to secure access everywhere that data is headed — with requisite compliance controls at every step. To succeed, companies need to think not about how to control access and how to mitigate risk, through techniques like masking and redaction, so they can provide easy access without fear of a critical breach.
Accomplishing this enterprise wide may feel like trying to boil the ocean, but it doesn’t have to be. Pick one or two critical datasets that you want to leverage for critical AI initiatives, and simply map out the end-to-end lifecycle. What is data gathered? Where is data propagated? How is risk mitigated? How quickly can consumers get access to the data they need?
As you start to answer these questions, you’ll likely find that investing in common data management platform can have a major impact. There’s no one-size-fits-all solution, but assembling the right technology foundation for your enterprise will massively accelerate your AI initiatives and associated business outcomes.
Access to data is a critical step, but all the data in the world won’t help if you don’t have the right people collaborating together to drive the AI outcomes you desire. To start with, people need to trust the data and each other. If your data is inaccurate or incomplete, or there is a lack of trust between organizations, it’s all too easy to undermine any data-driven conclusion with your own subjective opinion.
But we also need to understand whether we’re asking the right questions and whether the conclusions make sense. While this is part of any data-driven approach, it gets much more complicated in the world of AI. While there used to be a human that could explain their approach and why they did the things they did, AI introduces a black box machine into the process that breaks that chain of human understanding.
Instead, teams need to make sure they are bringing together a diverse set of skills to solve any data problem:
Someone who understands the structure and nuances of the underlying data, including where to find it if the right data is not at hand.
Someone who understands the business context, including what the goals are and what outcomes would be most impactful.
Someone who understands how the AI algorithms and machine learning models work.
Someone who understands how to apply the AI algorithms and machine learning models.
In some cases you can find all those skills in one person, but more often than not, you have to work proactively to bring the right people together. For example, you might run a clustering to understand how buying patterns are correlated with customer location. But is that important to the business? Do we actually have the right data to represent customer buying patterns? Is the data accurate, clean, and representative? Have we tuned the algorithm to do the right analysis for our particular dataset?
Trusting the black box of AI without an appropriate level of introspection is at best a waste of time, and at worst a source of catastrophic business outcomes. Invest in the required skills, and make sure that every data initiative brings those skills together in a collaborative environment to apply critical thinking to the problem at hand.
Artificial intelligence is upending the status quo of every industry. Whether you’re bank, retail store, engineer, or farmer - AI is changing how you do your job. While data management practices can look complex and intricate, data-driven companies today are ones that have the ability and competency to deliver fast, high quality, secure data so that data teams can effectively collaborate and innovate. Empowered teams with the right data and tools will be able to create new digital services, improve the customer experience, and drive better outcomes for the business.