Amii and Microsoft Canada Discuss AI Skill Shortage

August 31, 2022
AI Meetup

Amii held their virtual meetup in July and the theme of the discussion was around the AI skill shortage in Alberta, and the various career paths available for those interested in entering the industry. Hosted by Spencer Murray (Director, Communications and Public Relations at Amii) and featuring Joseph Pedrola (Product Owner – Training at Amii) and Warren Johnston (Product Owner – Talent at Amii), Spencer guided the discussion based on a whitepaper produced in collaboration between Microsoft Canada and the Alberta Institute of Machine Intelligence (Amii) through a series of ten interviews with executives and other technical leads across a spectrum of industries in Alberta that make up 42% of Alberta’s GDP, including agriculture, health, energy, operations & manufacturing.

The competitive advantage promised by AI is alluring for companies seeking to position their business for long-term profitability. These advantages include trend prediction in forecasting, decision making to maximize profits and minimize costs, and anomaly detection to identify unusual events to improve safety or reduce costs. As in any nascent and growing industry, the demand for the talent required to do the work outstrips the available supply. In a 2018 McKinsey report on barriers to AI adoption, 42% of respondents indicated that the lack of talent with appropriate skillsets being a barrier to adoption, the top barrier second to the lack of a clear AI strategy. Therefore, a company that hopes to adopt AI must have a well-articulated strategy to build their AI talent pool in order to be competitive.

Barriers to AI Adoption

The AI Adoption Spectrum is a tool developed by Amii to assess a company’s AI readiness and progress and consists of five stages:

Amii's AI Adoption Spectrum

Most companies interviewed are in the exploring and initiating stage, which means that they are just beginning to assess how AI can fit into their business and evaluate the resources required to implement AI solutions. Many companies usually start with using AI to streamline operations internally and companies further down the adoption spectrum will more frequently use AI in their client-facing product offerings. Similar to a company evaluating a software solution, the buy vs build question always arises with the buy option being the less expensive and less customized option, while the build option being the more expensive and more customized option. It is tempting to think that one can bypass all the AI talent management challenges with buying a solution, however, a rudimentary understanding of AI and what it can do for the business is still required to evaluate vendors and consultants, in addition to the post-implementation support of the solution. And if a company opts to build an AI solution in-house, then management would require the needed AI talent knowledge to recruit and assess candidates.

While one may imagine that the ability to turn data into an AI model to get some prediction results is all that’s required to implement AI, it is much more than that in actuality. For AI projects to be successful, one must understand the business domain thoroughly, and must work with the technical team and the business to gather the requirements needed to develop and implement the solution. Then once the solution is developed, business KPIs are needed to evaluate the success of the implementation. Therefore, not only is there a shortage currently in technical AI skills, such as advanced mathematics, programming, and data engineering, there is a large AI business skills gap shortage also, such as identifying what AI can and cannot do, translating between the technical and business teams, and managing stakeholders’ expectations, which is a vitally important skill given that a well-articulated communication to the right stakeholder means getting the greenlight for a project, and not being able to communicate clearly the value proposition means that the proposed AI project would not go ahead.

Therefore, one of the misconceptions with an AI career path is that one needs to have a Comp Sci degree to work in AI but that cannot be further from the truth as there are many opportunities in AI that are non-technical, such as product management, sales, marketing, data governance, AI ethics, data privacy and communications. The demand for AI and ML expertise is growing across all industries - and so is the demand for skilled workers. The Alberta Machine Intelligence Institute's Machine Learning Technician Certification Course gives individuals the skills they need to thrive as they pursue a career in AI or integrate ML into their current role. Part on-demand lessons, part live online classes and part collaborative project, the 9-week course is a fast and flexible way to develop your ML knowledge and skills. You'll leave the course with an electronic ML Technician Badge to share on Linkedin and beyond, as well as a Certificate of Completion and Capstone Project to share with prospective employers. Norquest College’s 4-term Machine Learning Analyst diploma program is also another great entry point to a career in AI that touches on every aspect of an AI project lifecycle.

There are generally three routes a company can go to acquire the needed AI skills. Hiring an AI specialist is the quickest way to do so, but it can be difficult given the AI talent shortage, especially with experts recommending starting with a senior-level hire that can provide both the technical skill and business leadership for an organization to gain initial traction. A formal skills development program, such as sending employees to a professional training program and conference so the employee can bring the knowledge back to the company is another way. Ad-hoc learning where employees work on a project and take as-needed training, such as a Coursera programs in reinforcement learning or NLP, to supplement missing skills is also another option.

Ways to Build up an Organizational AI Talent Base

If one is interested in a career in AI, it may seem overwhelming with all the available information on the internet. But like learning anything new in life, it’s important to find that mentor or someone who is in a position where you want to get to, whether it’s through LinkedIn, hackathons, or conferences and start asking questions as to where to begin and the steps you can take to get to that person’s position. You don’t need to know everything as the AI world moves so rapidly. Something that you learn today may not be as relevant five years from now, so the best skill you should learn is not any particular programming language, AI/ML framework or algorithm, but the ability to learn and adapt.

Despite of the enormous benefits AI can unleash, a clear execution strategy is required for success. A company must think about the various AI use cases for the business, and evaluate whether to buy or build, with each alternative affecting the company’s AI talent management. Technical and business skills are both required to succeed in an AI adoption strategy and a company must continuously assess skill gaps and implement strategies to fill these gaps. Motivational speaker Eric Thomas said that “at the end of pain, is success”, and while the pains of AI adoption and learning can be unbearable, the rewards are available at the finish line for companies and individuals willing to put in the work.

More from the YYC Data Blog