![](https://i2.wp.com/www.sapientmetrics.com/wp-content/uploads/2019/03/AI-Infographic.png?fit=740%2C284)
Getting Your Head around AI and Where to Start
Time to Read: ~11 minutes
As business technology evolves it is beginning to mimic the efficient data processing and decision making of the human brain. Just as the brain’s neural networks efficiently work through data to shape our thoughts, feelings, and behaviors, artificial intelligence is being shaped from the availability and capture of large volumes of business data. In combination with expansive computing power in the cloud and improved data science and analytics, teams continue to evolve and conflate their data, technology and business strategies. Like the systems of the human brain, the whole is more impactful than the sum of the parts.
Human intelligence relies heavily on the brain’s decision making. Some sources suggest that the average person makes an eye-popping 35,000 choices per day. After time for sleep, that totals about 2,000 decisions per hour or one decision every couple of seconds.
With such a high volume of data and decisions, the brain handles most of the work via the efficient subconscious. It is our subconscious, for example, that decides what data and information we onboard from our senses. Economics Nobel Prize winning psychologist Daniel Kahneman describes it as the dominant “System 1”, which makes decisions based in large part on past experience. If you see a tiger in your driveway, your subconscious will quickly recognize the threat as you don’t live near a zoo and dangerous wildlife is never seen on your block. These thoughts form in micro-seconds. Well before your conscious (“System 2”) is engaged, your emotions will already have alerted you to a threat. This highly efficient set-up is necessary to process the massive amount of data exposed to your senses; however, most decisions are processed without conscious interaction.
In your company’s environment, your data informs and represents the collective business intelligence of your firm. As such, the label “business intelligence” is as accurate as the label “artificial intelligence”. The difference begins in how decisions are made with this data: with or without human interaction. Similar to bypassing conscious reasoning, AI allows for the evolution of business decisions without direct human interaction.
The result of a business evolution driven by AI is a fundamental change in business decision management. Like moving decisions to the efficient subconcious, many actions will be automated in your data center. Post AI-transformation, insights are no longer based solely on deductive reasoning, but are augmented or controled by once seemingly disparate data stores and processes. The result is an orchestrated business intelligence – integrated, data driven insights and actions across your environment.
Where to Start – Data Evolution
If you don’t have a long-term data investment strategy in place, you’ve found your starting point on the path to AI transformation readiness. A robust data foundation is required to inform all intelligent platforms from process, to robotic, to business context. It’s not just about having the right data, but about managing it to support business goals and regulatory or security compliance. Collaborate with IT on a long-term data vision that includes strong infrastructure governance. The strategy will likely evolve over time and you can still proceed in other digital investments, but this is the lynchpin to long-term success in your AI transformation.
To see why your data evolution is so important, let’s look ahead a few steps to when machine learning algorithms will likely be leveraged. Sarah Butcher at efinancialCareers describes several types of machine learning (emphasis mine):
The purpose of supervised learning is to establish a relationship between two datasets and to use one dataset to forecast the other. The purpose of unsupervised learning is to try to understand the structure of data and to identify the main drivers behind it. The purpose of deep learning is to use multi-layered neural networks to analyze a trend, while reinforcement learning encourages algorithms to explore and find the most profitable strategies.
With these definitions, you can start to see the power behind your data and why the quality of your interdependent data stores is so important.
The graphic below shows the physical architecture components and coordination necessary for machine learning functionality. The machine learning code is represented by the small black box in the middle. Every other box represents the framework necessary to support such analysis. To trust a forecast, you must be confident in the data from which it is curated. Having trust in your data foundation becomes exponentially more important as your data dependencies evolve.
For an orchestrated use of, and visibility into the breadth of systems and data stores necessary for your AI evolution, be sure your IT governance stratgegy is in line with your long-term business goals AND budgeted for success. As one CEO framed it, “As a rule of thumb, you can expect the transition of your enterprise company to ML will be about 100X harder than your transition to mobile.”
Step 2- Metrics and Team
Susan Ettinger, an analyst with Altimeter Group, correctly reasons “If you move to a more rules-based, hierarchical world, to a more probabilistic, scientific world…getting your team right is kind of critically important”.
Unfortunately, business teams are rarely in tune with the metrics being driven from above and are equally unaware of many metrics in their own day-to-day. According to an expansive decade long study published in the Harvard Business Review, “Only 55% of the middle managers can name even one of their company’s top five priorities.” “Fewer than one-third of senior executives’ direct reports clearly understand the connections between corporate priorities, and the share plummets to 16% for frontline supervisors and team leaders.”
In short, do not assume your normal channels for communicating transformation objectives are effective. Additionally, be sure your vision defines how data insights may be leveraged and how they relate to your corporate strategies. Communicating your vision is necessary to not only address uncertainty, but to ready the team to identify new AI opportunities and quantify their decisions.
There are many consultants ready to assist in steering your plans (my favorite is Dr. Edward Peters) but, as with your communication strategy, start by re-examining your current assumptions. A first-principles approach combined with solid root-cause analysis in your metrics will provide a great map for where to best focus your resources, as well as how and what to communicate to your team.
Finally, it is necessary to consider the short-term as well as long-term ROI variables. Most teams wield technology solutions to improve efficiencies. What if data agility is also a long-term goal? This scenario may warrant a reduced short-term ROI in one project, in order to achieve your 5- year objectives. Ian Barkin, Chief Strategy and Marketing Officer with Sykes published this recommended training, AI, RPA and Cognitive Tech for Leaders. In it, he speaks to a common set of metrics and references the “return on platform”, taking into consideration the long-term goals across your environment, noting that the whole can be greater than the sum of the parts.
Unfortunately, knowing how much we will benefit from a platform of orchestrated data intelligence can be like trying to quantify “happiness”. If quantifying your hard and soft metrics is a challenge, I highly recommend the trainings and book offered by Douglas Hubbard, “How to Measure Anything” for your leadership team. His approach is grounded in the same statistical reasoning of AI, and his trainings are an investment that will continue to pay off well after your evolution.
Moving up the stack with RPA
As you research the many possibilities and technologies espousing AI competencies, you may find several of your current vendors already incorporate data-driven insights into their tools. But what is a good next step for your own data stack?
To answer this question, let’s look at a big picture perspective on AI transformation captured in this infographic. You have likely already started your evolution at the first level by digitizing different business systems with tools like customer relationship management (CRM) and resource planning software like SAP. Though the AI-stack shown here is not meant to represent an exact linear progression, it provides a logical perspective of moving your business and technology teams towards the data-driven, actionable insights you seek. It also highlights an important goal of orchestrated business intelligence: visibility to real-time, data-driven insights with the ability to take automated actions across your environment.
Several analysts propose Robotic Process Automation (RPA), seen in the automation layer above, as a main driver in an organization’s digital transformation. RPA software integrates with your existing environment through the same software interfaces used by your employees as it mimics human actions in handling repetitive processes. From manual data entry to routine tasks, RPA bots work through these processes with immense speed and accuracy.
Many teams find healthy benefits by automating existing processes with RPA to reduce errors, increase throughput and become familiar with the technology. However, RPA is not just about optimizing processes to be more efficient. It is particularly suited to integration tasks where employees provide a bridge to disparate systems and data sets. An important note: the goal is not simply to automate, but to re-engineer processes across silos to support and inform your long-term data vision. In addition to the potential cost savings of automated processes, RPA improves data collection and analysis providing improved standardization, compliance and detection of anomalies.
Moving forward, your team can explore more advanced AI-based automation, for example, using visual and cognitive intelligence that draws information from multiple sources to recognize patterns across your data stores and predict future events. Each step is part of an evolution towards an environment of integrated processes and data stores that provide for orchestrated business intelligence.
If you don’t shoot, you can’t play
According to a report from Forrester on the state of digital business: “Digital is approaching a tipping point. Over the next five years, companies will begin to see digital affect the majority of their revenues. Most of today’s companies are unprepared for this change”.
Most analysts would agree, the transformation isn’t optional. In order to stay in the game, digital transformations are table-stakes. Ignoring the revolution won’t just leave you behind, it will exclude your team from being able to play at all.
With this spotlight, look for where decisions are being made based on fairly consistent set of circumstances or that are dependent on manual data entry. If those decisions were handled or resolved automatically with 97+% accuracy and a measurable increase in CSAT, could you measure the benefits? Then grab your team and discover how to transform your business with insights you never expected!
Think big, start small and learn fast! – Good Luck!