True AI readiness must go far beyond the data, and empower (and reassure) the people responsible for its success.
How is your organization preparing for artificial intelligence (AI)? Ask this question of businesses investing in this field today, and the answer almost always comes down to “data”– with leaders talking about “data preparations” or “data science talent acquisition.”
While there would be no AI without data, enterprises that fail to ready the other side of the equation– people– don’t just stunt their capacity for good AI, they risk sunk investment and jeopardize employee trust, brand backlash or worse.
After all, people are the ones building, measuring, consuming and determining the success of AI in enterprise and consumer settings. They’re the ones whose jobs will change; whose tedium will be eased by automation; whose consumption or rejection of AI’s outcomes will be the focus.
People, in short, are those who’ll feel AI’s myriad impacts. That’s why investing in AI is as much about investing in people as it is data.
I wanted to dig deeper into this issue. So, my co-founders and other industry analysts at Kaleido Insights and I surveyed more than 25 businesses that have deployed AI at scale to learn about the ways they’ve invested in people. Here is what we found:
1. Investment in factors beyond technical talent
Hiring a team of data scientists will not cause business processes to magically become automated overnight. Some liken this mistaken assumption to hiring electrical engineers to run a bakery: While the mechanics of ovens are important, it is the experienced baker who best knows how to innovate recipes and inspire customer delight!
Across industries, we found that the successful AI deployments we saw involved at least eight distinct personae:
- Product leaders
- Front-line associates (e.g., customer support agents, field technicians)
- Subject matter experts (e.g., doctors, security admins, legal, etc.)
- End users
- Data scientists & technical builders
In addition to identifying these stakeholders, businesses have to make AI accessible and build trust by educating people and quelling fears. The top recommendation here is to prepare stakeholders by using tactics that put AI into context for each role.
Leadership requires a demonstration of ROI and visualization. AI leaders at FedEx, for example, built simulated dashboards and reports to illustrate the difference between traditional analytics and machine-learning-driven recommendations.
Meanwhile, readying the sales team requires both equipping agents with the knowledge, tools and confidence to sell the benefits of AI, and re-evaluating their metrics and incentive models to preserve quality and integrity. For effective roll-out, the unique needs and pain points for each of the above staff members have to be addressed.
2. Investment in addressing AI’s cultural stigma
AI is distinct from other technologies in that it can challenge people’s sense of importance and relevance. Some 58 percent of organizations in international settings have not discussed AI’s impact on the workforce with employees, according to a recent survey by the Workforce Institute. Yet AI’s success is driven by people’s willingness to adopt it.
Thus, enterprises deploying AI are well advised to assess how people’s sentiments, fears, questions and insecurities impact their proclivity to adopt. Instead of ignoring concerns, companies interviewed suggested discussing and developing positions and initiatives to address:
- Job displacement
- Algorithmic bias
- Privacy, surveillance
- Security threats
- Autonomous machines
- Societal manipulation
- Environmental impacts
- The notion of “killer robots”
These “elephants in the room” don’t just threaten employee morale, they highlight opportunities for companies to improve engagement and reinforce a healthy and trustworthy company culture. Address concerns of job displacement at your own company by evangelizing the limitations of AI. Articulate where AI will augment or accelerate human workflows. Provide clarity on governance models. And support employee upskilling and continued education programs.
Microsoft’s Professional Program for AI is an example: This is a massive open online course (MOOC) designed to guide aspiring AI builders through a range of topics, from statistics to ethics to research design. Other companies, like Starbucks and Kaiser Permanente, have partnered with elearning platforms like Coursera or Linda.com to facilitate professional development.
3. Investment in building an AI mindset
While investing in a mindset might sound squishy or disconnected from the bottom line, preparing employees with the education, ownership, tools and processes they need to engage with AI has tangible business benefits. According to a recent survey of 1,075 companies in 12 industries, the more companies embraced active employee involvement in AI design and deployment, the better their AI initiatives performed in terms of speed, cost savings, revenues and other operational measures.
The following “3 D’s” of what I call the AI mindset reflect three universal truths about AI and serve as starting points for building people’s engagement in an organization’s AI journey:
Think “diversified”: AI must be designed and managed by multiple skill sets. Those responsible for the day-to-day administration of the workflow are the ones who best understand where the breakdowns occur, where products fall short, where they, the staffers, spend most of their time and where customer sensitivities lie.
The business benefits: Diversifying AI design and development helps companies identify important features, UX/UI needs and use cases that might otherwise go unseen, or take more resources to surface. Companies like Wells Fargo have cross-functional centers of excellence to accelerate this process, emphasizing the value of using trusted internal influencers to facilitate onboarding.
Think “directional”: AI implementation is not a linear, “completed” destination, but rather one that calls for continual learning and iterations based on feedback loops.
The business benefits: Instilling a “directional” mindset reduces time to at-scale deployment. Even though people want to see results quickly, the extent of experimentation determines how strong any AI model is, and how many problems it can solve. Often, deployment time is based on user adoption, and the more people who can help train and optimize the system, (again) the more problems adoption can solve. This is also why companies like SEB, a Swiss bank, deployed its virtual agent, Aida, to 600 employees; then to 15,000 employees, before rolling the agent out across its million-plus customers.
Think “democratized”: AI is more sustainable when organizations enable accessible tools, training and multi-functional contribution and collaboration.