Artificial intelligence (AI) development has grown rapidly in recent years as business, governments and organizations search for new insights and efficiencies. With that growth comes new opportunities for employment in this sometimes-misunderstood field.
Even amid the rising flood of COVID-19 news, AI captured headlines: AI for tracking the spread of infectious diseases, AI for sifting through tens of thousands of coronavirus-related research papers for clues to a new vaccine and AI at hospital call centres for triaging patients potentially suffering from COVID-19 as opposed to less-threatening ailments.
AI seems to be everywhere, but what does this work look like? What are the qualifications needed? What is the demand for this skillset now and in the future?
General versus narrow AI
In drawing a portrait of an AI worker, AI’s role in the fight against the coronavirus pandemic serves as an outline: healthcare subject experts working with computer experts to develop new tools to improve health outcomes.
“When people talk about AI, they’re often thinking about general AI – effectively a machine with a capacity for human-like cognition,” says Dr Peter Taillon, Senior Data Analyst with the data team at the Information and Communications Technology Council (ICTC). “Arguably, services such as IBM Watson try to sell themselves as that, but it’s a place we’re nowhere near yet. What we are getting really good at is machine learning.”
Taillon, who holds a PhD in Computer Science from Carleton University, expands on the distinction between general and narrow AI. Narrow AI is commonly referred to as machine learning, which uses historical data and statistical techniques to create models that can then be used for predictive analytics, for example. This is what recommender engines at Amazon or Netflix do: they analyze your behavioural patterns and make suggestions based on your historical choices.
“AI seems to be everywhere, but what does this work look like? What are the qualifications needed?”
In industry, machine learning projects usually pair software developers and subject-matter experts in everyday jobs. While there are some companies dedicated just to AI development, most AI work today takes place within specific industry sectors across the economy.
“Essentially, an AI developer is somebody who’s writing software to solve a problem in a specific domain,” Taillon says. “For example, if you work in oil and gas, then your applications of AI are very specific. You’re hired by British Petroleum to improve the reliability of production and refining, based on historical data and other specific modelling parametres – it’s business demands driving AI solutions.”
Effective AI solutions are born from the coupling of technical software development skills and deep sector understanding. These two skillsets are seldom found in one person. To be effective, developers need to be good at working with other people – AI development is typically a team sport. That means AI developers need to be good collaborators, which goes against the lonesome computer-nerd stereotype.
“You need to be able to bridge that technology/sector expertise gap to explain highly technical models and techniques to basically very smart business laypeople and vice versa,” Taillon says.
A software developer assimilates domain knowledge to build the software platform. This requires technical skills: solid mathematical and algorithms knowledge, and a fluency with computer languages. Depending on who you talk to, the most prevalent languages in AI development are Python, R, Java, C or C++. Each has strengths depending on the application.
One factor that drives the acceleration of AI development today is that developers don’t need to write machine learning code from scratch. They can draw upon libraries of already developed programming.
The use of prepackaged material doesn’t necessarily soften the technical skills requirement, but it does make things faster and, again, hints at the collaborative dimension of AI development.
“As with any human knowledge, we stand on the shoulders of giants – in this case, giant nerds,” Taillon says. He laughs when he says that – he doesn’t mean any disrespect. In fact, he’s part of same cohort, with a proficiency in at least five programming languages.
Now and in the future
The demand for AI is spreading across the economy, and educational institutions are responding with course offerings that formally bring together technology and domain expertise.
“You now have degrees in computational economics and computational biology. There’s a whole slew of courses and programs that are ‘computational’ something,” Taillon says. “What they’re doing is taking a domain and explicitly incorporating the computer science and machine learning components.”
As AI development progresses, he anticipates that narrow AI will continue to have high impact, as you won’t see pure, general AI any time soon. Meanwhile, the current “insanely high expectations” for technical skills posted in AI job ads will likely become more grounded.
“A lot of these ads call for master’s or PhD of computer science or mathematics. Then you look at the job description and say, ‘Come on, you don’t need that.’ That’s overkill for a lot of this work,” Taillon says.
Dr Peter J. Taillon is a Senior Data Analyst with the Data Team at the Information and Communications Technology Council (ICTC). He has extensive experience spanning both the academic and private sector, having worked as a professor, researcher, management consultant and software developer with SMEs and start-ups. His expertise is in artificial intelligence, data science, Internet-of-Things/sensor networks and Big Data.
Want the best of CareerWise delivered to your inbox each week? Subscribe to our popular CareerWise Weekly newsletter to receive top news and views in career development every Tuesday.