We’ve all heard stories about facial recognition software that misses dark-skinned persons or Robo-loan officers that refuse to lend to certain populations. According to a growing body of evidence, algorithms designed by non-representative groups have resulted in AI that exacerbates the inequalities that already exist in our society. This algorithmic discrimination of data literacy concern is expected to intensify as more organizations rely on data and AI.
Most firms are already aware of this. They’re trying to figure out: how they can prevent becoming yet another bad example?
The simple answer is that everyone’s responsibility should be to think critically about the data they acquire and how they use it. The only way to develop ethical AI is to broaden the circle of people in the room. The one who contribute to question, constructing, and monitoring algorithms. That work necessitates data literacy. This is the capacity to process and organize complicated data, evaluate and synthesize information, generate forecasts. It also understands the ethical implications of algorithms. They are frequently more practical than academic.
Building data literacy in an organization can also help to diversify the data teams that make critical decisions regarding data acquisition, analysis, and dissemination. As a quant fund manager for more than a decade, I saw the advantages of varied data teams firsthand. It is commonly considered that portfolios with a wider range of investments outperform because they are less risky. Diverse teams, on the other hand, outperform because groupthink is reduced. By investing in data literacy across the organization, businesses may bring more diverse and creative perspectives to bear on both avoiding the risk of algorithmic bias and identifying extra efficiencies and opportunities that data can frequently reveal.
However, statistics show that the majority of businesses are still attempting to enhance their data literacy. Only 25% of employees are confident in their data abilities, even though 95% of company leaders feel data literacy is critical to corporate performance. Furthermore, according to some estimates, nearly nine out of ten data science experts are Caucasian, with just 18% being female. In terms of diversity, data science lags behind even other tech-oriented sectors such as digital marketing and user experience design, according to a General Assembly survey.
Why aren’t we teaching data literacy consistently and at scale, despite the clear need and growing urgency? For the past few years, that has been the driving force behind my work. My team works with financial services organizations and Fortune 500 corporations to establish more inclusive pipelines of data science talent at Correlation One, which I co-founded after leaving my fund in 2018. Supporting companies ranging from Target to Johnson & Johnson to the Colombian government in evaluating their current workforce’s capabilities. They offer free training to aspiring data scientists (like our partnership with SoftBank and the city of Miami). We’ve had firsthand experience with the crucial need for a more data-literate workforce, and we’ve aided businesses in putting policies in place to make that goal a reality.
Here are some of the strategies we use.
Data literacy should be a top goal for the whole company, not just the IT department.
It is not a technical skill to be data literate. It’s a highly specialized field of knowledge. Host quarterly engagement sessions on topics like data-driven decision making, the art of the possible in AI, how data connects to your business, AI ethics, and how to communicate with data to encourage all of your employees — marketers, salespeople, operations personnel, product managers, and so on — to improve their data literacy. The foundation of a data-first cultural transformation is this sort of organizational-wide focus.
Create an internal standard language for talking about data, how it relates to your business and industry, and how it affects certain positions within your company.
The data world is enormous, filled with jargon and perplexity. Develop an organizational vision of which aspects of data literacy is most relevant to your business — for example, probability and risk measurement may be significant in a financial services firm; experimentation and visualization may be key in a technology firm. In your L&D sessions, provide learning content that uses this language and explains how it relates to your organization in various areas, so employees can connect the connections between data literacy and their processes.
Make areas inside your company for employees to link business and data principles.
We advise all of Correlation One’s clients to enable their staff to produce new business ideas based on their data literacy. Assume your firm is in the music sector, for example. As part of your L&D program, have workers create project proposals that use their increased data literacy skills. By combining it with their industry experience, they will offer startling new ideas for cost reductions or revenue growth. Equally essential, you will enable them to develop a new data-first culture from the ground up.
Create incentives to support data-driven decision-making.
Take a look at how you now accept ideas and create budgets. Include methods that enable data-driven thinking after that. Require managers to incorporate clear visualizations in their proposals or establish dashboards. They monitor their KPIs quantitatively and in real-time, for example. You’ll rapidly gain the behavior you want from your managers. If you can shift their decision-making from intuition to data by rewarding data-driven suggestions with faster project approvals or larger budgets.
Deploy L&D programs that educate data literacy in the context of your business concerns – and that engage your staff.
Subscriptions to education and training platforms such as Coursera frequently fall flat in firms seeking long-term transformation. Creating these personalized, social, and contextual learning programs requires additional resources. The advantages in terms of employee engagement with the content, retention of the information, and empowerment of your employees are well worth it.
Perhaps most significantly, my experience both before and during Correlation One taught me that data is not vertical. It is not limited to a single job family, such as data scientist or data engineer. Instead, data is horizontal. It is a set of abilities that can be used to a rising number of occupations in every industry. A marketer with data expertise is a better marketer. A product manager with data skills is a stronger product manager. And so it goes for operations, engineering, sales, and even human resources. Everyone doesn’t need to be able to code. However, data literacy will become a must for everyone shortly.
Finally, data literacy encompasses far more than machine learning and data science. And it’s about more than just AI. Data literacy is about humans surviving better in a data-driven environment. That is why we need it today more than ever.