Machine learning, a subset of artificial intelligence, can have a tremendous impact on an organization’s ability to make better decisions and deliver better product features to customers. With valuable use cases as wide-reaching as IT operational enhancements to image recognition, natural language processing to recommendation systems, companies are racing to ramp up their machine learning capabilities.
Machine learning is transforming the way routine tasks get done. The implementation of machine learning in the enterprise setting is freeing up the workforce to work in a more strategic way. But like any new business advantage enabled by technological advances, there are headwinds to negotiate in the implementation process. In the case of machine learning, companies are faced with a skills gap problem. To program, interact with, and fully exploit machines that learn, humans have to have specialized coding, communication and data science skills. Yet, a recent study shows that the majority of companies do not have a plan to expand employee skill sets in these ways.
Below are five steps that together can serve as your organization’s first maneuver in a training plan for machine learning success.
1) Understand the benefits
“Machine learning” is as potent a buzzword as “big data” was ten years ago. In many ways, machine learning is what makes the promise of big data realizable. But before machine learning can affect the way non-routine cognitive work gets done in the office, first a team must do the cognitive work of defining what the machines should do - and how much autonomy the machines will have.
In banking, for example, there’s technology that can pre-qualify someone for a loan faster and more accurately than a team of people used to, but there are still quality-assurance steps built in for humans to make the final decision. In healthcare, every bit of information about an illness ever recorded can be cross-referenced in seconds to make a precise diagnoses, but the role of the trained, experienced physician in making the final judgment call is still critical. The key for you and your business is to get a clear picture of what data you have, how clean and actionable it really is, and the risk-reward inherent in the choice to allow machines to make autonomous decisions within your business model.
2) Socialize the vision
Because there is an aspect of machine learning that involves the displacement of current jobs, there may be underlying uncertainty in your organization about the impact that your machine learning strategy will have. Get in front of this anxious feeling by sharing the positive vision of how this technology will be a great thing to the company and everyone involved. Here’s an example: Hack Reactor’s Enterprise Training team is working with a company who is having an annual tech summit where we’ve been invited to lead a machine learning workshop - a tremendous way to get dozens and dozens of internal influencers involved in the thinking behind machine learning and leading the conversation on how it could positively transform the business.
3) Brainstorm capability needs
What is it that your workforce will need to be doing to achieve business goals in regards to machine learning? What are the knowledge and skill areas that are most important to machine learning? Clarity in regards to these two questions will go a long way in helping you evaluate current job positions and if you have the right organizational design.
Consult with internal SMEs or external machine learning thought leaders to focus in on the critical tasks that must be performed on teams building machine learning products or managing machine learning capabilities. What does “great” look like on these teams? Documenting observations around this question can help in building a machine learning competency framework for how you want to develop machine learning skills in your company.
4) Assess current skills
Once you have a competency framework, develop scenario-based questions with good-better-best answers and have your team gauge their current skills against what’s needed. Don’t overcomplicate the exercise. You’re really looking for a broad-brushstroke picture of where the skills gaps on your team exist, which will inform future training decisions on how to address this situation in a way that best meets your needs.
Be transparent with employees that this exercise is a learning and development growth opportunity which will enable the company to identify the training resources needed to support machine learning competencies.
5) Develop a learning pathway
With all of the information you will have gathered in the above steps, you’ll be in a great position to develop trainings whose outcomes will deliver enhanced machine learning capability in your organization.
How big should your catalog be? Machine learning is broad subject. If you’re interested in image recognition or natural language processing, you’ll probably want to investigate how deep learning with convolutional neural networks can help you over your technical hurdles. If you’re interested in predicting future events or handwriting, you may want to look at applications of the long short-term memory features of recurrent neural networks. If you are modeling inexact information in a decision-making system, perhaps trainable fuzzy logic controllers can meet your needs. If you need to find optimal strategies for existing bottlenecks, you may want to consider some of the fastest forms of computational optimization such as simulated annealing or swarm intelligence algorithms.
Need help narrowing down the offering? Study your audience. Get a sense of where learners are now (just learning and in need of lots of oversight, or strong grasp of foundational material and in search of advanced topics). Get started with what’s needed today, then iterate and expand over time.
Hack Reactor Enterprise Training offers immersive courses in machine learning as part of our comprehensive suite of tech skills solutions, spanning Assessment, Onboarding, Reskilling, and Upskilling. Have an assessment or training need? Visit us at www.hackreactor.com/enterprise.