AI is poised to transform nearly every industry, and with it will come significant changes for many job functions. Many roles across organizations will require at least some use of artificial intelligence technologies in the coming years, creating massive new opportunities for the AI-savvy regardless of discipline.
Alongside this transformation of how many IT and business staff do their work will be the emergence of new jobs targeted at making the most of organizational AI strategies. Machine learning engineers have already cemented their place as must-have members of AI teams, taking first place in Indeed’s best jobs list last year. AI specialist was also the top job in LinkedIn’s 2020 Emerging Jobs report, with 74 percent annual growth in the last four years, followed by robots engineer and data scientist.
In fact, even during the pandemic, the number of AI-related jobs could increase globally by 13 to 16 percent, according to IDC analyst Ritu Jyoti.
“Because of the pandemic, IDC believes that AI spending and employment will increase among healthcare providers, education, insurance, pharmaceutical companies and federal governments,” she says.
We reached out to IT leaders, AI experts, and industry analysts to get a sense of the kinds of AI roles they see emerging as AI takes firmer hold of the enterprise. Some leading-edge companies are already filling these positions, lending insight into the mix of skills necessary to succeed in them.
Chief AI officer
AI leadership roles fall under an array of titles: vice president of AI and machine learning, chief innovation officer, chief digital officer, and many more.
Whatever the name, these “chief AI officers” must understand how cognitive technologies impact the busines, develop the company’s AI strategy, and explain it to the board, company executives, employees, and customers. And they work with the CIO to implement this strategy to best meet the needs of the business and all stakeholders.
Nicole Eagan, chief AI officer at cybersecurity firm Darktrace, splits her time working with in-house technology teams, talking to customers, and evangelizing the firm’s AI strategy, which includes figuring out how to augment human efforts with AI, for example, in detecting and investigating threats.
“I work with the CTO and our AI lab to explore new areas for research and development,” says Eagan, who previously worked up the ranks of Oracle’s strategy group to become senior director of strategic marketing.
Eagan continuously augment her technical AI skills through online classes, but her role at Darktrace is more business-focused, applying AI to real-world problems, rather than creating algorithms and writing code. “We do have over 35 PhDs with advanced math, machine learning and AI expertise who are working in our labs,” she says.
Howie Xu, vice president of AI and machine learning at Zscaler, rose up the technical ranks, supplementing his experience with business skills. The former head of Cisco’s cloud and networking services business unit has an MBA from Stanford and a deep product background.
“When I first joined Zscaler, my role was more about the technology,” he says. “But in order to take full advantage of AI and ML, I had to evolve to think more about business impact.”
He recommends chief AI aspirants focus on areas where AI and ML can bring a ten-fold business value improvement to the table. “Be disciplined about the business metrics before the technology,” he says.
AI ethics officer
AI ethics officer is another high-level position that requires extensive work with stakeholders. The role may also cover risk and governance, and may need to coordinate with government agencies, nonprofits, legal teams, users, and privacy groups, in addition to technology teams.
Kathy Baxter, architect for ethical AI practice at Salesforce.com, says AI ethics officers must have a passion for technology but also a healthy skepticism. “AI is not magic and is not appropriate for every challenge. You need to frequently ask not just ‘can we do this?’ but ‘should we do this?'” says Baxter, who previously worked at Google, eBay and Oracle in user experience research.
Although technical literacy is extremely helpful, AI ethics officers do not need to be computer scientists or data scientists, she says. “What is more important is to have a humanistic background like psychology, sociology, philosophy, or human-computer interaction,” she says. “It is critical to focus on understanding everyone impacted by technology, their needs, context, and values.”
Baxter, who has a masters in human factors engineering and a bachelor’s in applied psychology, also credits the ability to de-escalate emotional debates as helpful. “When we talk about ethics, people can feel like their values are being challenged,” she says. “Being able to have healthy debates in an inclusive manner can be the difference between success and failure.”
Companies that pay attention to ethics when deploying AI create safer, more just environments, she says. Moreover, unbiased AI is more accurate and leads to better business performance.
“AI regulation is coming so creating an ethical AI practice now will better prepare you to be in compliance,” she adds.
AI business analyst
To reap value from AI models, data scientists must be paired with business analysts, says Shuman Ghosemajumder, global head of artificial intelligence at Shape Security, who has already hired someone from this role, and will be expanding the area over time.
“An AI business analyst should have a strong understanding of the company, its business model, and the business processes or product they are hoping to develop AI solutions for,” he says, adding that they also need to speak tech to work with data scientists and data engineers.
A related role, AI business operations manager, works on the business side to manage and improve business processes that use AI. “An AI business operations manager should have foundational knowledge in operations and experience in the particular business processes which are being automated through AI,” Ghosemajumder says. They should also be able to analyze data generated by those operations.
Finding people to take on business-oriented AI roles may be harder than it sounds, says Anand Rao, partner and global AI leader at PricewaterhouseCoopers.
“Universities and other vocational training institutions are competing to train a number of entry-level technical jobs,” he says. “However, the business and executive jobs need to be grown and cultivated within the firm and will pose a significant challenge to fill.”
Chief data scientist
Typically the top technical AI job at a company, the chief data scientist role has been evolving to include more engineering and business skills.
“Data scientists five years ago were statisticians,” says Brian McCarthy, head of analytics transformation at McKinsey & Co. “What we’re seeing now is that data scientists come from more of a technology background.”
Data scientists know what data to use and what algorithms to deploy to get the best results, working with data engineers and software developers to turn this know-how into working applications — and with business units to ensure the technology meets business needs.
Michael Roytman, chief data scientist at Kenna Security, got his masters in operations research in 2012 at the Georgia Institute of Technology, where he studied stochastic processes and optimizations. He then signed on to be a data scientist at Kenna Security, where he eventually was promoted to chief data scientist.
“Chief data scientists are moving into applying their skills sets to enhance analytics capabilities across the organization,” he says.
AI architects — also known as AI or ML engineers — are responsible for creating the systems to operate and manage AI and ML projects.
“These are people who can look at AI projects at scale,” says Steve Whittaker, head of strategic U.S. academic research partnerships at BT, and the head of its research partnership with MIT. IT architects who acquire AI and ML skills are good candidates for these jobs, he says.
“If you’re building an AI engineer platform, you need DevOps skills,” he says. “You need to know how to execute at scale, understand agile development, and have a sense of process and data.”
AI architect may also be responsible for rebuilding business processes, thus aligning them closer to the business.
Any company building its own AI or ML infrastructure will need AI architects or AI platform engineers. “It’s not just the Googles, Facebooks, and Amazons,” he says, adding that the recency of the role means that backgrounds vary widely, from new grads with new ideas, to people with 40 years of practical project management experience.
eSentire CTO Dustin Hillard expects ML engineers to have several years of experience working with large data sets and cloud data processing frameworks, and have the ability to design, build, and deploy complex AI systems.
AI data engineer
Both AI and machine learning live and die on data. But the data required can differ in kind and scale from that needed by other systems, so any organization that wants to perform advanced analytics, ML or AI will need an AI data engineer.
“Large organizations are the natural thing that springs to mind,” says Kevin Brown, managing director of security at BT, of the kinds of companies that should be looking to hire for this emerging role. “But also other organizations that have massive data. Health care, for example, is seeing exponential growth in data as a result of the pandemic.”
At BT, for example, the quantity of data processed is staggering. On the cybersecurity side, for example, there are millions of events per second, and about 4,000 cyberattacks per day. The company employs a managing director focusing solely on AI and strategy, Brown says, as well as AI developers, researchers, data scientists — a broad range of AI functions.
“We have a vast amount of data that we quickly need to sift through to find the anomalies,” he says, and this is where AI data engineers come in. “We’re always looking for the needle in the haystack.”
Data manufacturing architect
Companies in the data business offer even more specialized roles. For example, Bloomberg recently hired someone to fill a new role of data manufacturing architect on its CTO Data Science team.
The data manufacturing architect helps Bloomberg create high-quality structured data for its financial services customers, including more than 325,000 Bloomberg Terminal clients. The data comes out of unstructured, noisy sources, says Gideon Mann, head of data science, in Bloomberg’s Office of the CTO.
“These numbers must be precise and accurate, with standards beyond those of most industries and academics,” he says.
The data manufacturing architect works deep domain specialists in Bloomberg’s global data department, he says. Bloomberg is also hiring for a number of other specialized AI jobs right now, including AI research scientist, AI quantitative research scientist, human computation architect, senior ML engineer for media data science, and senior software engineer for distributed systems.
These roles require experience in AL, ML, natural language processing, information retrieval and quantitative finance, says Anju Kambadur, head of AI engineering at Bloomberg, and they should have expertise in programming languages such as Python, Java and C++. But communication, collaboration and product development skills are also important, he adds, “in particular the ability to work and communicate across organizational boundaries and across disciplines.”
AI quality assurance manager
Additional AI-related jobs are emerging to fill the needs of bleeding-edge companies, as they try to figure out how to allocate responsibilities around nascent AI practices. Some of these jobs don’t really exist yet on the job market, and most won’t have a standardized curriculum or typical career development path.
Take, for example, the emerging role of AI quality assurance manager, which could be viewed as evolving from a traditional software quality assurance job, but quality assurance for AI projects is dramatically different. The code itself is rarely the issue, for example, although a company may choose the wrong algorithm for the project at hand. What is more important are incomplete, out of date, or biased training data sets.
Biased data is a particularly thorny problem that can lead not only to bad results, but also regulatory implications, bad publicity, fines, or lawsuits.
“Nobody really understands how bias gets into the data, and how to try to get it out of the data,” says John O’Neil, chief data scientist at Edgewise Networks, recently acquired by Zscaler. “It’s an active area of research. As far as I know, there’s not a place you can go and say, here are the rules, and if you follow the rules, you’ll be okay.”
Citizen data scientist
According to Gartner, by 2024, AI power users will fill the talent gap for data scientists. These “citizen data scientists,” as Gartner calls them, will be able to perform AI-related tasks because the tools needed to deploy advanced analytics, machine learning and artificial intelligence will become easier and easier to use.
Don’t look for this as a job title, however. Instead, experience with “citizen data scientist” tools, such as Auto ML, will be part of job descriptions for a range of job functions.
“The traditional data scientists are expensive to hire, scale and train,” says Ryohei Fujimaki, CEO and founder at DotData, an AI platform company.
But around 28 percent of AI and machine learning initiatives have failed, according to IDC’s March survey, in large part due to skills shortages. “Lack of staff with necessary expertise is reported as one of the primary reasons for the failure,” IDC’s Jyoti says.
That means that there is pent-up demand for reskilling of workers for AI and ML skills, she says.
And more and more of a need for “citizen data scientists,” says DotData’s Fujimaki.