Decoding complexity and the pivotal role of Artificial Intelligence

Complexity has traditionally been viewed as a negative disruption and a dominant rationale for why organizations do not achieve their strategic results.? Simplification has been an overarching alternative theme and is often identified as the proposed solution for the organization but is this still the right choice given the advancements and potential as we engage with AI? According to Nelson Cowan, as humans we can only hold 3 to 5 items in our minds at once (1), so we naturally gravitate to simple concepts.? This may further explain why, according to BCG, 70% of all complex transformations fail or do not live up to expectations (2).? Are we over simplifying our solutions and are not addressing the complex problem in front of us?? Is there a need to challenge our natural instinct to strive for simplicity because human behaviors, today’s systems, and ways of working are more complex than ever?? As an alternative, can we take the opportunity to embrace complexity and leverage AI to solve this rubik's cube???

What if we embraced a revised approach that challenges the status quo?? One that?promotes leveraging AI in a new way because it is built to handle complexity on our behalf. An approach that promotes (functional)?complexity, moves with speed (to avoid analysis/paralysis), and leverages data that is good enough but maybe not perfect.? In the world of AI, we can leverage parallel processing (over linear) and produce more output while staying aligned to these principles.? If we don’t, organizations miss the opportunity to enhance the accuracy and efficiency of decision making and to improve our ways of working.??????

First, we must rethink our perspective on the benefits of simplicity as a key to success and instead, embrace the complexity of our systems.? Don’t over complicate it and tackle one problem at a time, is often the mantra.? In today's technology ecosystem, we have numerous tools that overlap and complement each other.? In this quest to simplify, we reduce the power or capability the tools may be able to deliver and un-intentionally create data silos.? Per Salesforce, 70% of organizations don’t provide connected experiences for their end users(3).? As leaders, we push our teams to provide executive summary level information that is understandable, digestible and can be endorsed at an executive level.? This results in leaders understanding the problem statement and satisfaction that the summarized version of our solution will solve the challenge at hand.? We then require our solution architects to follow this same rule and cascade this approach to our teams. What if we change this mindset and still allow for simplicity with our executive leaders but then we embrace the complexity as we design the solution?? Complexity in this definition is allowing our solutions to tackle multiple problems, connect silos and provide a solution that leverages AI to generate and answer the next 5 questions about a problem and not just solve for a single use case or problem statement.? Take the example of matching a candidate to a job, an early adopted and strong example of successful use of AI in a business process.? Now imagine if that same matching capability could be leveraged to showcase gaps in skills, that then trigger learning recommendations for employees or plot a leader on a succession plan that is outside of their business unit or geography simply by asking the next question in the natural evolution -?rather than HR leaders painstakingly and inefficiently executing this activity manually and not at scale.? According to Ken Murgage, “the key challenge in software today is embracing complexity; not treating it as something to be minimized at all costs but a challenge that requires thoughtfulness in processes, practices and governance.”(4)??

Next, we must encourage the focus on ‘speed’ and progress, always evaluating time to value.? As we evaluate technology and how AI can play a role, we have all been warned about the perfect software implementation where everything works flawlessly at the outset.? We have seen the skills taxonomy, the job architecture, and use of position management all work in perfect harmony.? We have been conditioned to believe that all of this is required for our technology investment to pay off.? The counterpoint to this is that we can allow AI to tell us some of these answers and we can begin to leverage its capability and realize value to the organization now, instead of waiting for the perfect deployment.? The side projects around skills taxonomy, job architecture, etc.. will take time and possibly delay a decision or implementation, and therefore the path to value realization. ? A revised approach that embraces experimentation utilizing AI suggested outcomes allows us to move with much greater speed, realizing a new set of outcomes faster or even ones not thought possible.? Take the often used metric of revenue per employee.? Many business cases have this metric improving by 20%+ as a result of a transformation of people, process, technology and organizational structure.? This is one important outcome to validate any technology investment, but we must also look at the service level provided and the time it took to make a decision and then deploy.? By leveraging AI, we can do more with less and provide a new level of service to our candidates, employees, partners and customers in less time.? By embracing what AI can do, we allow the machine to work for us and allow us to spend our time in other places, realizing value that much sooner.? According to PWC Pulse Survey, 87% of CHROs are evaluating new ways to deliver HR value at lower costs and AI can be one way to solve for this (5).?

Finally, we must relook at our relationship with data.? In the past we have talked about systems of engagement versus systems of record.? The idea was that we interface with certain applications to perform transactions, allowing for the prioritization of user experience in these applications. Then, other systems maintain data in the background and prioritize its integrity and organization.? With data, the traditional thinking was based on the relationships of humans with this data - garbage in and garbage out.? We have not considered in this scenario, the power of AI to transform our data and infer things we could not see for ourselves (before our perfect transformation is executed).? In this old way of thinking, we limit AI to take what we provide as its only basis for interpretation.? The days of data dictionary projects are likely things of the past as we set AI on a course to interpret what we have and don’t, that is not constrained by our own human thinking.? Let's take the example of an internet search where we select the perfect words and leverage the proven boolean approach that returns a list of answers for us to then further research and select.? In a world where we leverage conversational AI, (ChatGPT Model 4 has been trained on 300 billion words) (6) we have an opportunity to interact with the technology to create a recommendation.? We begin by entering words and now those words are interpreted, contextualized, understood (not just taken at face value).? An answer is recommended that we have refined through interactions and don’t have to click on the recommended links, to then complete further research.? This example is one way in which we are embracing conversational? AI and allowing the AI applications to add value in places we maybe did not originally identify, that goes well beyond the data we originally provided.????

With adoption, AI has the power to transform the way we work.? This will be seen in places we expected and those we did not.? If AI is going to reach its potential we need to open the door but also embrace the role of the human with the machine as AI is differentiated by the depth it can apply versus having to select the best fit word search.? We have an opportunity to play a new and active role, are you ready??

Special Thank You to Ernest Ng, VP - Strategy Research at HiredScore for his contributions to this article.

References:?

1. Cowan, N. The Magical Mystery Four: How is Working Memory Capacity Limited, and Why? The Magical Mystery Four: How is Working Memory Capacity Limited, and Why? - PMC (nih.gov)
2. Patrick Forth, Tom Reichert, Romain de Laubier, and Saibal Chakraborty.? Flipping the odds on digital transformation success. ?What Employees Say About Agile Transformations | BCG
3. Mulesoft. MuleSoft’s 2022 Connectivity Benchmark Report.? 70% of Organizations Do Not Provide Completely Connected User Experiences, New MuleSoft Study Reveals - Salesforce
4. Ken Mugrage.? Why embracing complexity is the real challenge in software today.? Why embracing complexity is the real challenge in software today | Thoughtworks
5. PWC. ?PWC Pulse Survey, August 22,2023 CHRO and HR leader insights from the PwC Pulse Survey: PwC
6. Beatrice Nolan, Google researchers say they got OpenAI's ChatGPT to reveal some of its training data with just one word.? Google Researchers Got ChatGPT to Reveal Its Training Data, Study (businessinsider.com)