The question about Data Science that everyone’s asking

Will today’s hottest career become obsolete in 20 years?

The world is estimated to need 1 million data scientists by 2018, with an expected supply-demand gap of 40 to 50 percent. It’s not a stretch to say that data science is the most in-demand career there is right now.

And yet, optimism about the industry is often followed by worries about its future obsolescence. I meet young people who are afraid to take the plunge into data science because of concerns that machine learning and artificial intelligence (AI) will render them jobless in 20 years.

Fortunately, most of those worries are unfounded.

Why? Put simply, because the human brain is a fascinatingly complex organ. Scientists still don’t know how it works — they don’t know how thoughts are formed; why the same catalyzing event can spark very different trains of thought; or how humans make connections between seemingly random ideas to create entirely new ones. The human brain is what makes us unique and difficult to predict.

These mysteries are also why robots will never be able to imitate our ability to think. Enough behavioral data may eventually empower a machine to predict what an average human would do in a given situation, but it’ll never be able to articulate why that particular decision was the best one.

The ability to think critically, then, will be one of most important skills for the data scientist of the future. To compete successfully with a robot, he must be able to ask the right questions. He must be able to draw on a multitude of experiences and their inferences, and communicate those ideas effectively.

Instead, most young data scientists today focus almost entirely on technical skills: becoming experts in R and Python, being at the cutting edge of machine learning, and mastering the latest platforms. But in doing so, they ignore less tangible — but arguably just as important — critical reasoning skills.

The Limitations of Artificial Intelligence

Here’s what we know today about the shortcomings of AI:

1. Dependence on external input: The history of human thought is long, wide, and deep. We are the sum of our own lived experiences, as well as of humanity’s collective consciousness. Upon spotting an unusual object in the sky, for example, a human would use proxies like movies and books to infer that the object might be a UFO. But a neural network would not be able to make that connection. It simply won’t have access to that same wealth of data. There just isn’t enough processing power for a computer to hold all of humanity’s cultural references. There probably won’t be for a very long time.
2. Lack of contextual creation: We’ve gotten to the point where machines can now create new data. Generative adversarial networks, for example, are able to generate new data like video game scenery, but can only do so based on previous data. They can’t proxy data from other sources, or make the spontaneous connections that enable creativity. This limits their usefulness.
3. Open-ended problem-solving: As noted above, machines can only answer a question or solve a problem if they have the right kind of data to reference. Sometimes figuring out the answer to a complex problem requires asking the right questions. A machine won’t be able to creatively solve problems unless it already has a past precedent to rely on.
4. The need for clean data: Neural networks must be fed clean, well-labeled data in order to be trained for a new task. Consider the word lemon, for example. By most definitions, it is a yellow citrus fruit with high acidity. But the word lemon can also be used to mean the color yellow, a faulty product, or anime-based fan-work. AI can only discern the different meanings of the word when labels have been provided ahead of time.

The Data Scientist’s Secret Weapon

It might sound counter intuitive, but a data scientist’s future success relies on her ability to recognize — and utilize — her own unique identity.

In other words, realize that your particular amalgamation of experiences and thoughts makes your perspective invaluable, and use that liberally through the course of your career.

In the same vein, recognize that all the people you interact with come from vastly different cognitive frameworks. Understand individual motivations and needs, and use empathy to communicate with people.

Here are a few granular steps toward future-proofing your career. You’ll probably think of more as you progress but these will create a strong foundation.

1. Build your skills in statistical modeling and data cleaning: If the data is clean, robots become great tools for a data scientist’s work. But in the real world, data is never clean. One needs to have the highly-nuanced skills necessary to correct the information and develop appropriate models.
2. Become proficient at problem-solving: Become adept at translating the C-suite’s problems into solutions that leverage data. Become a great communicator — people in business units can’t always clearly articulate what they need. Seek to be embedded into business units rather than on a dedicated data science team so that your knowledge and offerings remain relevant to the business. Since naming the problem accurately is half the battle, let this be an area you excel in.
3. Increase your domain knowledge: Becoming a subject matter expert — say in healthcare or in finance — will enable you to connect seemingly random dots of information to form new features and develop unique insights.
   
   Take, for example, a situation in which data scientists are trying find new ways to prevent cancer. It’s already known that smokers have a higher rate of cancer. We also know that those who drink copious amounts of alcohol have a higher risk of cancer too. But what about the people who do both? Is the incidence rate of cancer higher with combined risks? A data scientist with expertise in disease prevention would immediately think to ask that question, but a machine would not.
4. Stay heavily involved with the business: Right now, data scientists tend to spend a lot of time with their heads down. But they instead need to show interest in solving business problems even outside the confines of their immediate project.
5. Stay well-informed: Be well read, keep abreast of the news, and use that knowledge to make smarter recommendations.

Remember: AI is ultimately just a tool and you are the master craftsman wielding it.

Springboard prepares you for the careers of tomorrow with its mentor-led online courses. Some courses even come with a job guarantee! Find out more.

Matt Fornito is a mentor for Springboard’s Data Science programs.