Artificial intelligence is an unavoidable concept in the digital age, if not for its pervasive application, then by virtue of the fact that it provides a technological solution to information overload and the quest for context. Enterprise interest in AI has been compounded by its promises for efficiency in the form of speed, accuracy, agility and access to insights embedded in “dark data” -- a reference to the estimated 80% of unused enterprise data.
Despite the surge in AI investment and intrigue, companies are struggling to know where to begin. To the extent companies are readying themselves for AI deployment, preparedness and investments therein center around data and data science talent. But AI demands much more.
Today, just one out of every three enterprise AI projects are succeeding. Companies may be experimenting with AI, but less than 30 percent have any kind of AI strategy in place, according to MIT Sloan. Virtually every company will attest to woefully inadequate data warehousing, storage or processing systems. Others recount cautionary tales of reticent, fearful employees or AI that failed to account for basic end-user needs. And even the largest AI companies in the world are encountering profound questions around ethics. True AI readiness requires more than preparing data, it requires readying the broader organization, chiefly people, process and principles.
Our research across some 27 businesses that have deployed AI at scale finds there are fundamental areas businesses must prepare for sustainable AI.
AI-driven transformation begins with ground-up problem-solving but must be aligned with broader data strategy and “digital transformation” objectives. While most AI initiatives emerge from within single business functions, organizational structure (particularly around AI governance) is the essential enabler of success -- for short-term wins and over time. While approaches and metrics vary by organizational maturity, measuring AI’s success must go beyond measuring financial impacts, and be prepared to shift over time.
The most practical area enterprises must ready for AI is of course, data. (No data, no AI, after all.) But data preparedness is not a linear destination. Companies must address the broader “data story” (from sourcing to storage, from stewardship to strategy, security and beyond) to bridge the gap between ideas and execution at scale. While many “ground-up” AI efforts prep data pipelines for single initiatives, our research finds enterprise AI has unique strategic requirements that are less linear and more about preparing data for an ongoing feedback loop of learning across applications, people and the organization.
Decision-making around the technical architecture and integrations required to deploy AI must align with core product strategy. Companies must balance reliability with flexibility and account for rapidly evolving capabilities across the entire “AI stack.” More often than not, deployments involve configurations of existing, off-the-shelf, open source and emerging software, hardware and data warehousing solutions. It is therefore essential configurations work toward reusability and avoid exasperating current IT and data inefficiencies.
The mass automation of big data and AI call for a new business competency: a formalized approach to ethics. Our research finds ethical preparedness falls into three broad categories. First, addressing bias -- human, data and algorithmic -- within and without. Second, driving transparency; AI introduces new levels of threats and complexity around which additional clarity is a strategic requirement. Third, the organization itself: Ethics officers, teams and even AI-tools are all emerging to address trade-offs, create guidelines and processes, diversify teams and develop best practices for cultivating ethics.
Preparing people for AI is as important as preparing data. Humans are, after all, the ones designing, deploying, measuring, adopting, rejecting and defining the value of AI. In practice, this means prioritizing human readiness over technological capabilities and addressing AI’s limitations and cultural stigma head-on. This requires instilling a distinct “AI Mindset” across myriad stakeholder groups, identifying key personae (our research identifies eight) and readying each accordingly to foster lockstep coordination between technical and product groups.
The very basis of AI -- understanding and reproducing human cognition -- renders it unique compared to other technologies facing businesses. It’s not just that it requires colossal (good) data to deliver accurate and actionable results. Or that it marks a shift in software development from deterministic to probabilistic. It’s not just that it can challenge people’s sense of importance and relevance, it’s that mimicking human cognition renders AI subject to increased scrutiny, over-inflated expectations and introspection. As the digital world grows evermore automated, companies have a unique role -- not just in extending AI’s practical and commercial applications, but in readying the broader organization for defining our relationship with it.
Jessica Groopman is an industry analyst and founding partner of Kaleido Insights.