In this special guest feature, Brian D’alessandro, Director of Data Science at SparkBeyond, discusses how AI is a learning curve, and exploring opportunities within the technology further extends its potential to enable transformation and generate impact. It can shape workflows to drive efficiency and growth opportunities, while automating other workflows and create new business models. Brian is also an adjunct professor for NYU’s Center for Data Science graduate degree program. Prior to SparkBeyond, he has built and led data science programs for several NYC tech startups, including Zocdoc and Dstillery. A veteran data scientist and leader with over 15 years of experience developing machine learning-driven practices and products, Brian holds several patents and has published dozens of peer-reviewed articles on the subjects of causal inference, large-scale machine learning, and data science ethics.
Despite all the data we have available, and the desire to leverage it with modern AI systems, humans are still heavily embedded in almost all business decision-making. Artificial intelligence can certainly help us understand our data and glean insights, but it is up to us to translate those insights into impact. Creating and investing in an AI strategy alone will not create competitive advantage for an organization. Business leaders need to always be committed to turn good models and insights into action, but will also need to trust these insights to do so.
Data alone has little intrinsic value. It is through action, and more specifically, how data changes action in value generating ways, that data reveals its worth. To ensure that an AI investment generates impact, organizations should create clear roadmaps on how to operationalize AI prototypes. An organization’s leaders should be engaged to ensure that each stakeholder business unit is working together with shared AI driven goals. Such AI adoption and support grants an organization a competitive edge to which it would be unquestionably harder for laggards to catch up.
Aligning the organization, acquiring data and committing to action are all necessary to create impact. But more can be done to drive further advantage.
Humans own the best tool to solve problems and come up with brilliant ideas: our brains. But this tool is only able to explore so many solution paths, and we often find that the questions businesses ask are those that are driven by a pre-existing hunch or intuition.
It is well known that the human brain has computational limits, but human ideation (and creativity) also has its limits. Our brains are designed by evolution to simplify the world and to only explore neural connections that conform to our past experience. This cognitive bias has helped us survive as mammals, but it could limit our growth in the age of AI. We are seeing a new wave of emerging technology designed specifically to tackle the bias inherent in human thinking — and thereby overcoming the cognitive bottleneck in reaching that eureka moment. Game-changing initiatives require a new way of thinking, a new way of generating ideas. AI’s impartiality and potential for idea generation is instrumental in getting us there.
Too often, AI is described as a “black box” — providing little transparency or ownership necessary for data decisions. This can lead to models that are hard to interpret, or insights whose basis are difficult to understand. Realistically, human oversight will be a part of most AI driven systems. Regardless of whether one chalks this up to human bias or responsible model governance, it is a reality that AI practitioners must incorporate into their development efforts. If AI impact is driven by the set of actions that AI enables, then earning the trust and confidence of all stakeholders involved is a critical component of AI success. Transparency should thus be a development goal. When AI is presented as a glass box, individuals at all levels of ownership and responsibility are empowered to use and trust the results in their decision making.
The world is dynamic and in constant progress. In order to genuinely affect change and address the root causes of business pain points, models need to continuously adapt to evolving conditions. These models aren’t only built to interpret the sheer volume and variety of data produced every day, but also harness its velocity. And that’s the precise advantage of using AI.
Models inherently become less accurate over time. If the training data isn’t refreshed often, the model becomes irrelevant. Employing AI when building models from the ground up enables an adaptive ‘model factory’ that allows us to retrain the model on refreshed drivers and data sets. As part of their AI strategy, organizations need to design and build systems that can adapt to a constantly shifting signal. Instead of thinking in terms of an “AI model,” teams should think in terms of an “AI modeling system.” Incorporating adaptability from the ground up ensures that any competitive advantage build off of data never gets stale.
AI is a learning curve, and exploring opportunities within the technology further extends its potential to enable transformation and generate impact. It can shape workflows to drive efficiency and growth opportunities, while automating other workflows and create new business models. While AI empowers us with the ability to predict the future — we have the opportunity to change it.
Sign up for the free insideBIGDATA newsletter.