Dive Brief:
- Organizations with an interdisciplinary team have a "far higher ratio of success" when deploying AI projects, said Arun Chandrasekaran, distinguished VP analyst at Gartner, speaking at a Gartner IT Symposium/Xpo Americas session last week.
- Interdisciplinary teams that blend roles across business and data science have a higher ratio of success with AI projects, as well as a faster time to production. This trend "clearly tells us that AI needs to be a team sport, said Chandrasekaran.
- "However, in reality what we see in most organizations is data scientists wearing too many hats, because there's a dearth of skills across other areas," he said.
Dive Insight:
The role AI plays in the organizational tech stack changed after the pandemic hit, its influence moving closer to key business processes such as upholding supply chains or enabling remote work.
But to sustain the AI momentum, businesses need access to the kind of talent that can simultaneously keep systems running and innovating. A mix of strategies can make companies reach the workforce they need as AI becomes more strategic.
Within a high-performing AI team, organizations need a set of clearly defined roles working in unison, according to Chandrasekaran:
- Data engineers: They're responsible for all aspects of data, from its ingestion and integration to the labeling of data and feature engineering.
- Data scientists: Leveraging insight from stakeholders, these experts design AI models to demonstrate critical use cases.
- Business domain experts: In conjunction with the data science teams, they work to identify use cases with high business impact.
- AI engineers: Typically trained in DevOps and software engineering, AI engineers deploy models in a production environment, providing a continuous feedback loop.
- Product managers: These experts help embed AI into core products or build uniquely differentiated AI projects.
- AI architects: This set of experts think about how all technologies blend with each other, and help guide core architectural decisions such as public cloud vs. hybrid cloud or build vs. buy.
Though talent acquisition and retention woes are common in tech writ-large, the dearth of skill in AI is glaring. C-suite members cite talent shortages among the key barriers to enterprise AI adoption.
"For every open position that's out there for an AI engineer, there are less than 15 people" that fit the criteria, Chandrasekaran said. This ratio includes employed people, which means there's even less talent available.
In search of scarce talent, organizations are wise to start from within, according to Chandrasekaran. But there are caveats to that approach.
"You have to take a very pragmatic view that these people will require a lot of training, upskilling and mentorship," in order to "become more versatile within the organization," he said.
To address industrywide tech talent shortages, companies such as Amazon have launched upskilling initiatives aimed at staff. IBM also launched a certification program focused on data science.
When hiring externally, tech leaders should focus on ensuring there's a culture fit with the incoming talent and the core values of the company, he said. They'll also need to be open to mentoring and tutoring existing staff.
Partnering with higher education organizations can supplement the strategy, though these tend to be longer-term strategies, while hiring external consultants can mean a high sticker price.
The majority of organizations, or 55%, take a hybrid approach that combines internal and external talent strategies in AI, according to Gartner data.
Talent in the AI space knows it's in high demand. Ultimately the ideal strategy includes explaining why an organization is the ideal place for an AI technologist. In crafting this reverse pitch, organizations must articulate a promise of great use cases, the ability to innovate or potential for growth.
"You have to really invoke your inner Steve Jobs, in some sense, to inspire these people to join your organization," said Chandrasekaran.
