Companies need professionals who have domain expertise, augmented with AI skills.
Many professionals and students want to jump into the world of AI. But rather than abandoning your current career track to become a data scientist or a machine learning engineer, consider developing AI skills to complement your existing subject matter expertise.
Employees with AI skills are in high demand right now. But can you imagine how much harder it is for employers to find someone who understands both AI and genomics, AI and physics, or AI and psychology? Let’s call these emerging careers “AI+X”.
Companies are investing millions to hire AI+X individuals and to train their subject-matter experts in AI. Scientists, engineers, and analysts from various disciplines have strong computational and coding abilities that allow them to quickly pick up the basics of AI: machine learning, data science, and/or software engineering skills. Those who combine foundations of AI skills with their own domain expertise are going to be well-equipped to become leaders across many industries.
Through Workera and DeepLearning.AI, I have been fortunate to meet and work with AI leaders from Fortune 500 companies. I learned that most companies look for individuals with dual competencies, including these recent listings:
AI+X individuals can hit the ground running, enabling companies to tackle new business opportunities more rapidly. Training a biotech engineer in AI could take months, but training an AI practitioner to understand biotechnologies could take years before meaningful output. A subject matter expert who comes with, or can develop, AI skills is a much better investment.
As an AI+X professional, you are uniquely qualified to build application-specific models even if you aren’t an AI expert. You understand the historical and technical rationale for various decisions, but also will be mindful of the limitations of machine learning. AI experts will help you make those models available to a wider audience.
In medicine, if you were to build a model to classify whether a patient needs a type of high-risk cancer surgery, the model might have an accuracy of 95%. As a data scientist, you can’t tell whether that number is good or bad. But if you are also someone who understands oncology you would understand the implications of a 5% false positive/negative result for patients, as well as the problems faced by a hospital or doctor who orders unnecessary operations. You would know which metrics and outcomes are acceptable and where the model must be further optimized. This nuance is a critical part of building and deploying successful AI models and an instinct that can only be tuned with domain expertise.
At Stanford University, Andrew Ng and I are already seeing this desire for obtaining an AI toolbox to complement existing knowledge. In our deep learning class (CS230), students learn about theoretical methods and also work on a project with hands-on mentorship from the teaching staff and industry partners. Two-thirds of our students are from majors outside the computer science department — everything from chemical engineering to astrophysics to law. In fact, non-CS students perform at par and often better than those pursuing a computer science degree, provided they have sufficient background in foundational skills. Many of them win project awards and some even have published their results in leading industry journals. A few notable examples:
- Tyler Quill, Shayta Roy and Yaakov Tuchman from the Department of Materials Science & Engineering predicted the melting point and viscosity of ionic liquids based on the component anion and cation chemical structures.
- Lily Buechler from the Department of Mechanical Engineering developed a deep learning framework to approximate the outputs from a power flow simulation, and evaluate performance for a variety of power network characteristics.
- Veronica Peng of Computer Science, Xi Yu from the Department of Bioengineering and Wenxi Zhao from the Department of Civil and Environmental Engineering used deep learning to classify gestures from divers communicating with an autonomous robot companion in dangerous underwater environments.
Our teaching staff tried to understand why non-CS students were so effective at delivering high quality AI projects and found three main reasons:
- Subject-matter expertise. This allows the non-CS students to understand their data and interpret results rapidly. To visualize how critical this is, imagine a non-Hindi speaker trying to build a Hindi text-to-speech synthesizer. It’s virtually impossible. Yet Dinesh Chaudhary did it in 10 weeks because he’s a native speaker.
- Access to data. They have quicker access to application-specific data than CS students through their network. Students from outside departments get incredible datasets from their labs that can be used to train AI models. Similarly in the professional world, if you are a structural engineer for an oil and gas company, you can have an impact in the field that a machine learning engineer at a tech firm never could.
- Time and Passion. These students tend to spend more time on their projects because it is not just a novel approach to their field of choice and often ground-breaking research but because they love what they do.
By receiving the right training, subject-matter experts can be effective at solving real-world problems with AI in a matter of months. In a business context, this can be what allows a company to identify a trend in their data before running out of capital or before slower competitors.
By building up your AI skills to support your subject-matter expertise, you will be able to:
- Spot when a problem is suited for an AI solution.
- Solve real-world problems with AI.
- Get an edge in the job market.
- Keep doing what you love while learning new technologies.
If you’re looking for inspiration and collaborators for an AI+X project, this list of deep learning projects and the Workera Slack group are good places to start. If you want to see how your AI capabilities stack up, take the Workera assessments, you might be surprised at how much you already know.
The AI software market is expected to make $126 billion in incremental revenue over the next 5 years. AI+X will be the largest driver of that explosive growth. Will you be part of the change? 🙂