Data Literacy is Not Optional

By Collin West

There is a well-documented skills mismatch in the United States today.

Each year millions of students graduate from colleges and universities, but are they prepared for the future of work in the 21st century?

According to Code.org, there are 70,000 computer science graduates each year but over 500,000 open positions in the United States. And the problem gets worse when you look at data literacy, the ability to understand and communicate the value from data.

“Artificial intelligence and data science are the biggest trends in history,” shares Rasheed Sabar, Co-Founder and Co-CEO of Correlation One. “The problem AI poses is not algorithms – it’s talent.”

The common thought is that artificial intelligence (AI) will lead to automation, and ultimately, job loss. However, AI will also create new jobs and change the nature of work for millions of existing jobs. The consensus view is that data is a vertical. People think of roles including data infrastructure, warehousing, visualization, analysis, and more.

But this view is incomplete and misses the truly disruptive nature of the data revolution.

Data is not a vertical; it is a horizontal category that touches every position across every industry. As Sham Mustafa, Correlation One’s Co-Founder and Co-CEO, said: “You can empower executives, marketers, sales reps, product managers, and more with data. Every job function, from the boardroom to the customer help desk, will require data literacy.”

The Current AI Challenges

For AI to reach its potential, we need to (i) train more talent and (ii) have people with diverse backgrounds and opinions build artificial intelligence systems. The talent needs to include engineers, developers, and researchers – but also extends to data-literate functional leaders in corporations, nonprofits, and governments.

We are far away from both goals. Today, up to 85% of the analytics workforce is male. And, even for leading tech firms, less than 5% of analytics professionals are black or Hispanic. To improve these outcomes, we have to look at the root cause of the problem: Why are there barriers to becoming data-literate?

Ensemble VC, together with Correlation One, has identified three barriers. 

First, the cost of modern education and training is too high. Either you have to pay $30,000 for a bootcamp, or twice that amount for a master’s degree. Online content, like massive open online courses (MOOCs), is good but has very low retention and engagement metrics.

Then there is the problem of social capital. According to Yahoo, up to 50% of job offers come from referrals, and referrals who get an interview are 40 percent more likely to be hired than other candidates. This creates an unfair playing field, favoring those who went to the right school or have the right friends and family.

Lastly, underrepresented minorities and women tend to lack mentorship. People can get the formal training that they need but need guidance on creating compelling resumes, preparing for job interviews, and negotiating higher salaries.

There are structural reasons that more people, particularly those with diverse backgrounds, do not get the education they need. We cannot ignore this problem any longer.

Data Science For All / Empowerment

Data Science For All / Empowerment is a program started by Correlation One. Their mission is powerful: to create equal access to the jobs of tomorrow. Their program provides instructor-led, practical training in data & analytics to 10,000 underrepresented minorities across the country, for free. 

For their inaugural cohort, they received 8,500 applicants over an 8-week period. Of that pool, they selected 500 participants. 

The talent pool was 57% female, 60% Black, 30% Latinx, and 13% LGBTQ. Other underrepresented groups include military veterans, athletes, and refugees – with representation from 48 states across the nation.

The program flips the education business model on its head, keeping the program 100% free to participants – with merit-based entry – and having corporate sponsors underwrite fellowships and scholarships for students. Sponsors recoup ROI by hiring program graduates, adding analytics capability, and building more diverse workforces.

“To get a digital transformation, you need a data transformation,” stated Rasheed Sabar. “History itself has inequities so if you naively look back at historical data to make decisions about the future algorithmically, you will perpetuate those biases. You need a nuanced understanding.”

In another program, Correlation One and SoftBank Group put together a 12-week intensive program to identify and train data science talent in Latin America. Their Data Science For All / Latin America (DS4A / LatAm) program boasted incredible results.

One team from the program developed a package routing algorithm that will save a SoftBank portfolio company an estimated $17 million U.S. dollars per year!

As a whole, DS4A / LatAm boasted 97% student retention and 100% project completion rates, which are unheard-of numbers for a program of this type.

Conclusion

Our current higher education system does not work for today’s workforce.

It’s archaic to ask someone to take 4-6 years to learn and never be formally retrained again. Not to mention the huge burden of higher costs, whether it’s borne by a government or an individual student.

This problem is not going away – and it needs to be solved. “You need to be re-skilled every 2-4 years,” said Sham Mustafa. “You cannot let people get trained once and expect that to last their entire career. They will get left behind.”

It’s hard to imagine a single interaction, from applying for a job or mortgage to making high-level policy decisions, that will not be reshaped with data.

We need a workforce of talented engineers and data scientists. That is undeniable. But, beyond technical skills, a modern workforce also needs to understand how to engage with data.

Data-literate executives, marketers, salespeople, product managers, and more will define the future of work. Questions like “What is this chart telling me?” and “How do I find the root cause?” can only be answered by someone who is comfortable with data.