Smarter Data Science. Cole Stryker
6, “Addressing Operational Disciplines on the AI Ladder”) should be viewed as a requisite offering that begins in grade school and must become part of the new-collar worker's continual learning within the workplace.
Building the organization's collective skills must encompass education in the use of collaborative software tools and socially oriented communication tools. Through being connected, employees can see who needs help and who can provide help, what problems need to be addressed, and how problems have been resolved. In democratizing data, organizations should notice that speed and value are moving in a positive direction, because sharing skills and knowledge can improve mutual understanding and business performance.
The impact of democratizing data and AI will circle back to refine existing job roles and responsibilities. Data scientists and citizen data scientists alike should be able to access and understand the curated datasets that are most relevant to support their own job functions. In building a workforce that is enabled to be data-driven through democratization, a new-collar workforce emerges. Organizations are faced with an unknown unknown in that this poses a new way to work but for which optimal organization structures have not been established. Change is upon the organization, but how that change manifests is not self-evident ahead of time. The new organizational structure of the enterprise is going to require frequent tuning.
Whether data science is applied by the data scientist or the citizen data scientist, sufficient oversight is necessary to ensure outcomes are not biased against the objectives of the organization. By empowering employees with essential skills, organizations can expand upon the opportunity to innovate and to find the next point of leverage. Sufficient oversight is also a concept that is distinct from sufficient insight. Sufficient insight would help to explain or articulate the what, how, where, who, when, and why of a singular outcome, whereas sufficient oversight would be the means to address causality across a series of outcomes.
Figure 2-4 shows the entwinement between the democratization of data and AI with data literacy and the ability to self-serve. The intersections should promote organizational collaboration, empowerment, and enablement of individuals and teams. The overall result is outcome-based in that the time-to-value proposition realized by an organization should be progressive and ultimately fair to all constituents.
Figure 2-4: Data and AI democratization
DEMOCRATIZATION
Four critical elements for enabling data democratization are that the data for which a person or a machine is entitled to see should be:
Easy to find
Understandable
Consumable
Of sufficient quality
For the most part, being easy to find means that you'll need to unilaterally catalog (or inventory) all of the data that exists within the enterprise and all of the applicable data that exists outside of the enterprise. The other elements are potentially nonunilateral in that understandability, consumability, and data quality are contextual and may vary for different people or different machines. For example, the names Kneel Fischman and Coal Striker may be of insufficient quality for the payroll department but be of sufficient quality for the internal fraud department.
Aye, a Prerequisite: Organizing Data Must Be a Forethought
NOTE
The word aye is British English for “yes.” The heading therefore reads as “Yes, a Prerequisite.” As a pun and wordplay, the pronunciation “eye, a prerequisite” sounds the same as “IA prerequisite” where IA are the initials for information architecture, a focus topic of this book. Therefore, the heading also reads: “information architecture prerequisite.”
All organizations, regardless of size and industry vertical, are actively engaged on a journey to Valhalla: a place of honor, glory, and sustained organizational happiness. This journey, with all of its twists and turns, involves the need to embrace analytics. The recent collective embrace of analytics stems from the observation that analytics has become the low-hanging fruit for addressing organizational change. The expression low-hanging fruit refers to tasks or actions that are most easily achieved. The analogy comes from the very literal task of picking fruit off a tree without the need to use a ladder. But organizations that believe they can retain their operational status quo run the risk of irrelevance or eventually obsolescence. For many organizations, analytics is now a vehicle for helping organizations figure out what can be done now and what can be done next.
Although AI and enabling technologies can carry a higher degree of panache, many of the foundational skills required to fully deliver on AI's promise are not well honed. These fundamentals of information architecture aim to address the problem of deriving value from data, which is inherently inert, not self-aware, and not self-organizing. Information architecture addresses these characteristics of data and aims to organize and contextualize data so that it can be mined for value.
NOTE
An enabling technology is an innovation that can be leveraged to drive radical change in the performance or the capabilities of an organization. The Internet and wireless communication are examples of enabling technologies.
Having an expectation that AI can consistently work its magic on any data source—regardless of the type of data and regardless of the level of data quality—without having a proper information architecture is a form of naivety that is all too widespread. Information architecture is the prerequisite to maximizing the benefit of advanced analytics, especially neural nets, machine learning, natural language processing, and other forms of AI.
Organizational leaders understand that change is constant, accelerating, and arriving from all sides. In the United States, the 1933 Glass-Steagall Act forbade commercial banks from owning securities firms.
NOTE
For more information, read Eric Weiner's book What Goes Up: The Uncensored History of Modern Wall Street as Told by the Bankers, Brokers, CEOs, and Scoundrels Who Made It Happen (New York, NY: Back Bay Books, 2005).
Seventy-five years later, the securities industry all but came to an end when the remaining large players, Morgan Stanley and Goldman Sachs, requested permission to become bank holding companies. Modern organizations must be willing to recognize that change can happen at any time and that when something changes, that something is also susceptible to being changed later. While the time horizon between changes can vary from seconds to years and decades, placing a corporate anchor on something that proves volatile can be disastrous.
NOTE
For more information about the changes at Morgan Stanley and Goldman Sachs, read Ben White's article in the New York Times, “Starting a New Era at Goldman and Morgan” (September 22, 2008).
Competitive companies tend not to remain stagnant, especially those that formally declare through a mission or vision statement the desire to improve shareholder value. By restructuring the organization, acquiring another organization, spinning off a line of business, or growing organically, organizations continually change and evolve. Companies that change in ways beyond an existing market niche can certainly reap assistance through the use of AI. Paradoxically, if an overarching strategy is to tightly align information technology with the business, that may be a surefire way to stifle or inhibit business growth and prevent the ability to rapidly change.
An