Guiding the change management side of AI with SCENE: A personality framework for AI transformations

GAIn® introduces SCENE as the first personality framework originating in the world of Artificial Intelligence, built to let organizations succeed in their AI transformation by proactively overcoming the challenges of human change management.

Artificial Intelligence (AI) is omnipresent in our daily lives and continues proving its business value for enterprises. Other than with previous revolutionary innovations – such as the steam engine and automobile – AI is replacing the decision-making process of humans. As was the case with those previously emerging technologies, a one-on-one replacement will not result in the full potential of AI. Therefore, businesses should build processes where humans and machines collaborate.

However, transitioning to such a collaboration is not easy. The largest challenge does not lie on the machine side, but on the human side, since AI fundamentally changes the human expert way of working and redefines their role as humans. To better understand the human-side of this equation, one should start with understanding the personalities of those involved in an AI transformation. Adjusting to the expert’s willingness to change and what will trigger this same willingness, as well as promoting diverse personalities within project teams will foster AI adoption.

Consequently, a suitable personality test is needed. Although many personality tests are available, lots of them lack scientific evidence and a data-driven approach. The Big Five personality framework does possess these qualities, yet the absence of a typology – like for example the MBTI personality types – makes it difficult to use in a business context. Therefore, GAIn® developed SCENE: the first personality framework that finds its origin within the field of Artificial Intelligence, where the Big Five – backed with its extensive research – is well-translated into the field of AI. Through analyses and extensive domain expertise, eight key profiles are identified that all contribute in their own way towards a successful adoption of AI. SCENE is aimed at making AI transformations work for enterprises through understanding human personalities. In this article we introduce SCENE, discuss the need for an AI personality framework, the SCENE typologies, and how SCENE helps organizations succeed with AI.

The need for a scientific personality framework to successfully guide AI transformations

Artificial Intelligence is changing life as we know it. According to Google CEO Sundar Pichai, AI is one of the biggest developments in human history. Bigger than fire and electricity. We are interacting with algorithms on a daily basis, whether we are looking something up on Google, or searching for the next series to binge on Netflix. Moreover, in organizations AI is no longer a futuristic notion, it is only used by a few Data Scientists tucked away in the corner of the office building. As Paul Daugherty and James Wilson describe in their book Human + Machine: Reimagining Work in the Age of AI, AI is transforming our way of working and redefining our role as humans.

According to Daugherty and Wilson, much value is expected to come from the collaboration between humans and algorithms, defined as Augmented Intelligence by Gartner. Gartner estimates the value created by AI augmentation in 2021 at $2.9 trillion of business value and 6.2 billion hours of worker productivity globally. (1)

So how do experts adopt AI? Different personalities have a different trigger for change. Experts who are akin to visionaries are moved by the innovative character of AI, whilst guardians are more protective towards the status quo and are inclined to choose proven solutions. Additionally, philosophers that are introvert thinkers would first like to understand the complete mechanism of the human-machine interaction before they are willing to adopt it.

Furthermore, on the organizational dimension, personality plays a key role to deliver on the promise of value creation with Artificial Intelligence. Visionaries are needed, who dare to change, inspire, and motivate others for continuous innovation. Critical analyzers are necessary to concretize the business processes where AI can potentially make the largest impact. Controllers who are detail-oriented have a leading role in creating high-quality algorithms that consistently show high performance. And managers are crucial to create the multi-disciplinary teams needed and guide them to realistic targets.

 

Logo_GAIn Personality Framework

 

Alternative personality tests lack a data-driven approach in combination with a translation to the business world

To understand and optimally profit from the talents and mindsets within an organization, personality tests can give a lot of insight. However, although there are many personality tests available in the world, such as the Myers Briggs Type Indicator (MBTI) or Insights Discovery, many of them lack scientific evidence and a data-driven approach. Most of the popular personality frameworks are based on Carl Jung’s psychological theory. And although the theory is easily and widely applicable, it lacks scientific evidence for its assumptions.

The Big Five framework seems to be the most fitting, as it has earned its credibility and popularity due to its scientific validity as well as the robustness of its conclusions regardless of gender, culture, and so on. However, the Big Five is also criticized for its negative connotation on some personality traits and labeling them as good or bad. Whereas we foresee an important role for all personalities within AI transformations. Also, the absence of personality typologies within the Big Five decreases the practical suitability of this framework.

Therefore, to facilitate and support successful AI transformations, GAIn is launching SCENE – the GAIn® personality framework that is specifically tailored for use in an AI context. Based on analysis and extensive domain expertise, eight key profiles are identified that all contribute in their own way towards a successful adoption of AI.

An illustration of a human-machine collaboration within health

Yet, we have been quite abstract about the changing role of business experts, where they make decisions with algorithms, jointly. To make it more concrete, let us illustrate an example from the role of specialists in medical diagnostics. When a patient comes to the hospital with symptoms of asthma or COPD, several tests will be done, and a chest X-ray will be made. This X-ray results in an image that is sent to a pathologist who specializes in lung diseases and makes a diagnosis by looking at these images. However, with modern image recognition techniques, the classification of lung tissue and subsequent diagnostics can be enhanced by an AI algorithm, where the pathologist and the algorithm collaborate to make the most accurate diagnosis. In this new situation, only a selection of images for which the algorithm is too unsure will have to be judged by the pathologist. As a result, the algorithm does the frequent diagnoses accurately, while leaving the edge cases to the pathologist. This drastically changes the role of the pathologist. On the one side, the pathologist must trust the algorithm. On the other hand, he or she must understand when the algorithm is off, why it is off, and how to improve this. We see the same developments in businesses, where the role of business specialists is changing as more and more of their knowledge is accommodated in AI algorithms.

SCENE FIGURE

Introducing SCENE: how analytics have given rise to the five traits and the eight typologies

SCENE represents five fundamental personality traits as a two-way continuum, where the opposite ends represent the polar extremes of a particular dimension. Both with their unique qualities and pitfalls that make successful diverse teams.
SCENE scale-1
The SCENE Framework.
 
The GAIn® personality framework is a mean to describe an individual’s personality. More specifically, it is a hierarchical personality framework that measures a person’s character across five independent traits or facets: Social Interaction, Cognitive Style, Energy Source, Need for Control, and Emotional Reaction (SCENE). Each dimension in the SCENE framework measures a unique facet of a personality, and individuals are scored on all five. These five traits are – obviously – not all-encompassing of the complexity of the human mind. However, they illustrate the fundamental drivers behind human behavior and consequently allow us to explain and adjust behavioral patterns.
 
Note that the personality traits are clustered into three leadership traits and two self-management traits. Leadership traits amass to the question: What defines success for you and how do you want to reach it? Success itself is not uniform, even though we try to capture it through various key performance indicators (KPIs) in our normal work-life. For instance, we observe that cooperative people derive success based on consensus, whilst the critical end of the same trait derives success on the quality of the output. The two self-management traits determine what you need to excel in in the work that you do, especially under pressure. How do you organize yourself, and how do you cope when things do not go the way to wanted or expected them to go?
 
Based on the IPIP-NE0-120 items questionnaire, the resulting report provides an extensive personality profile that identifies and describes the strengths and pitfalls of individuals. Each trait is driven by six sub-facets, which gives participants a unique insight into each personality trait. For instance, extraversion is oftentimes perceived as the desire to be around other people and the ability to easily connect with others. However, extraversion can also be defined as the ability to be assertive and take the lead. During the many trainings we have already given with SCENE, we see that participants can understand their internal motives better through these sub-facets.

Eight SCENE typologies provide a detailed interpretation that is applicable in the world of AI

The five dimensions of SCENE are independent traits that together describe the personality of an individual. However, the traits do not all have an independent effect on our behavior and our personality, and subsequent behavior is driven by the interactions between our personality traits. This holds especially for the three leadership traits, which describe your objectives and interactions in (complex) social situations. These leadership dimensions therefore form the basis for eight distinctive segments, each with a unique contribution to the team. In the combination of their traits, unique talents, and pitfalls arise.
 
Note that each individual has several elements or characteristics from each type, but often one or two dominate the others and can be considered as an individual’s preferred type.

SCENE profiles-1

The eight SCENE typologies.

Confirmatory analytics as a foundation of the eight typologies

The eight SCENE typologies are constituted based on extensive expert knowledge, and publicly available research, and also through analysis on the results of over 200,000 respondents of the IPIP-NEO-120 test – an open source Big Five questionnaire. A cluster algorithm (k-nearest-neighbors) conducted on this dataset shows the existence of eleven personality clusters, where every individual belongs to one of these clusters. When mapping the eleven clusters on the eight SCENE profiles (see graph below) an interesting insight arises: On average 74% of the individuals in each cluster can be mapped on two SCENE profiles. Stated differently: 74% of the variance within a cluster is captured by the top two profiles.
 
What does this number of 74% say about the validity of the SCENE typologies? If there would be no relation between the eleven clusters that the algorithm found and the eight SCENE typologies, this same percentage would be 25%; one third of the actual value observed.
 
This analysis is core to the use and interpretation of the SCENE typologies. As the eight profiles are easy to interpret, and are collectively exhaustive for people’s leadership traits, use of these profiles is preferred over the use of the eleven clusters. However, we do treat the typologies as a preference instead of a fixed personality. Someone’s most dominant profile will be the most preferred in their mindset and behavior, but they have access to the talents (and pitfalls) of their other dominant profiles as well.

Confirmatory analysis-1

Mapping of the eleven clusters from a clustering algorithm on the eight SCENE typologies.

The human-side is the largest challenge within AI adoption, not the machine-side

As mentioned before, the largest impact within AI is created through building systems where humans and machines collaborate in a decision-making process. In the success of building such AI solutions, personal diversity plays a key role, for two reasons. First, creating impactful AI solutions asks for personality traits that are unlikely to be found within a single individual. Second, the adoption of AI within an organization requires a culture shift that all different personalities must accept and embrace different approaches towards this change.

When building AI solutions, multiple disciplines are needed: The business expert who will work in the decision-making process jointly with the algorithm, the Data Scientist building the algorithms, and the Data Engineer who guarantees the algorithm is fed by data and its results are shared with the business experts. Each of these roles have different responsibilities, and different personality types thrive better within certain roles. For instance, Data Engineers excel in a more deterministic environment where the solution must be extremely robust since it must operate effortlessly day in, day out. Whereas the work of a Data Scientist is less predictable since their field is about distinguishing what is certain and what is not, which can change unexpectedly in the near future. Apart from their respective expertises, a Data Engineer needs qualities such as pragmatism, an eye for detail, and discipline to excel in his or her job. Data Engineers are therefore oftentimes Guardians, Controllers, or Managers. On the other side, a Data Scientist must be curious, creative, and innovative to find the best model solution for the challenge at hand. More often, a Data Scientist will be a Visionary, Challenger, or Philosopher.

Moreover, in the adoption of AI within organizations, personality also plays an essential part. A survey by O’Reilly(2) shows that the number one bottleneck for AI adoption is the company culture that does not recognize the needs for AI. Drivers of culture are manifold, but usually the willingness to change starts from the top and cascades down to the rest of the company’s hierarchy, from where the business will pick up an idea and move to execution.

AI adoption bottleneck-1

Personality plays an important role whether and in what way people adopt a new way-of-working: Different personalities have a different tendency towards embracing new developments. Some are very open to new experiences and jump on new ideas (too) quickly – these are people with a high score on curious - whereas others first need to be convinced about the added value of a new approach over an existing approach – these are people with a tendency towards pragmatism.

Illustrating the power of SCENE in creating better algorithms

 

To illustrate the power of SCENE as a tool to nourish the success of AI transformations, we look at the process of developing impactful algorithms, which lies at the core of AI solutions. Typically, there are four challenges that AI encounter in the model creation process, that also show why working in multidisciplinary teams is so important.

  1. Creating a model that creates actual business value. In practice, modeling projects oftentimes fail because the model does not answer a relevant business question, or because the model is not trusted by the end-users. Building impactful models therefore requires deep understanding of the relevant business questions, and also strong communication with the end-users during the modeling process to ensure the adoption of the final model.

  2. Choosing the right model for the problem at hand. Building models requires deep technical knowledge of different modeling techniques and when to use which. Especially in the field of AI, that is evolving at an exponential rate, teams need to keep up with new developments continuously, but also understand when existing techniques would be the better option.

  3. Ethics and the law. Models are unbiased, they simply use the data they are fed with without judging its fairness. However, models based on personal or group characteristics might become unfair or even racist (and illegal) when they are not moderated. An example of this are models created by the Dutch Tax Authority to trace fraud, which discriminated against people with a dual nationality. (3) Moderation of models and the adoption of Responsible AI is therefore an important challenge that AI teams must overcome.

  4. Sustainable implementation of models in business processes. If an organization wants to adopt the use of algorithms at scale, they must be technically implemented in existing business processes. This requires the automation of data flows and model scoring, the development of user applications, and also the long-term monitoring of the model performance. This calls for the strong collaboration between Data Scientists, Data Engineers, and app developers, and strong communication with the business, but also the understanding when performance is still acceptable or when an intervention is necessary.

 

Diversity as a core quality for building algorithms

This list of challenges is not complete, but they already show that building models does not only require strong technical knowledge, but also a diversity of personalities within the team. From sharp Listeners who want to understand the business problem, to passionate Inspirers who move an organization to become more AI driven. From Innovators always seeking new techniques to Implementers who refactor code and fix bugs.

To overcome the challenges and build impactful AI algorithms, five personal qualities must be present in the team:
 
  • Having the quality of observing and listening to the needs of the business stakeholders, and proactively and passionately inspire others for change. (Energy source)

  • A positive-skeptical attitude, where the team challenges the results obtained by their team members, while there is also an atmosphere in the team in which team members respect each other’s opinions and solutions, and cooperate with each other to get to the best solution. (Social Interaction)

  • Showing result-oriented creativity, that enables a team to constantly look for innovative solutions to the problem at hand – for example in the creation of new features, or in the use of new modeling techniques – while at the same time keeping the focus on the end result. (Cognitive Style)

  • Being able to conceptualize problems into a model, by really understanding the business context and the dynamics of the event to be modeled. And in the next step, being able to translate this to a list of concrete data fields that are either existing or that must be created from the available data. (Cognitive Style)

  • Being flexible to change the strategy based on the outcome of the model, while working consistently and structured towards a realistic goal. (Need for Control)

Mapped onto the five dimensions of the SCENE personality framework, these five qualities hold contradicting personality traits. Look for example at the first quality. Being a sharp listener is a quality of an introverted personality, who oftentimes prefers observing group dynamics to voicing their own opinions. On the other hand, inspiring others for change demands a level of assertiveness and willingness to take charge. This is a typical trait of an extraverted personality. Both are necessary in the team, both are traits on the dimension of Energy Source, yet they are on the opposite ends of the scale and will likely not be prevalent in the same person.
 
Now let’s look at the second quality of a positive-skeptical attitude. The word itself already exists of two parts that seem to be each other’s opposites: skeptical, yet positive. On the one hand people who are willing to challenge others on their ideas, who stay objective, and who are willing to take the risk of being less likable for the benefit of the result. These are typically characteristics of people who score high on Critical on the dimension of Social Interaction. On the other hand, a team needs people who are concerned about the opinions of others, who are constructive, and connective. People who can mediate and get a team to a widely supported solution. These are archetypal skills of people who score high on Cooperative on that same dimension of Social Interaction. Both crucial skills for a high-impact team, yet again on opposite sides of the same dimension.
 
Obviously, one can continue this rationale for the other three qualities, emphasizing the need for diverse teams. And that is where the GAIn® Personality Test – that maps a personality on the five dimensions of SCENE – comes to play. Based on the extensive personal reports, SCENE provides insight into the strengths and pitfalls of the team and provides guidance how to gain a team benefit from individual talents. To build teams who can overcome the challenges and create real impact with AI.

 Experiencing SCENE in our GAIn® trainings

For people who want to experience the power of SCENE themselves or together with their team, GAIn® offers several badges on this topic in our Leadership & Change domain. These badges cover the topics of stakeholder management, team management, and effective communication. During the training you learn and experience how to use your own SCENE profile, and that of the people you are working or interacting with, to improve the effectiveness of your communication to reach your objectives.

For more information on the possibilities, browse our website.

Authors

This article was written by Lara Buimer, Bart Maassen, and Marcel Mol. All being part of the SCENE team that further develops the framework and aims at incorporating SCENE into GAIn's Leadership and Change badges. Please reach out to GAIn if you have any questions to us. We're happy to have a chat!

Do you want to assess the current skill gaps in your organization? Get in touch with Bart or Marcel to learn more about our company approach with SCENE.