UK investigates Facebook over data breach, to raid Cambridge Analytica – Reuters
Cambridge Analytica and a moral reckoning in silicon valley – The New Yorker
Cambridge Analytica and Facebook: The Scandal and the Fallout So Far – New York Times
These are some of the headlines from previous weeks amidst the public outrage and media frenzy regarding the misuse of Facebook data by the Big Data firm, Cambridge Analytica. To the shock and horror of many concerned citizens, the UK based firm allegedly tapped into the accounts of 50 million Facebook users, accessed their data without consent and subsequently “harvested” the data to influence an election result.
Using the power of and data science and behavioural psychology, Cambridge Analytica was able to develop psychological profiles of specific voters within the United States. These profiles – refined and calibrated by accessing Facebook’s massive data set were subsequently used to influence voters through the use of targeted messages within the Trump political campaign – a move which has shocked the political and tech world alike. For some, this was an eye-opening glimpse into the Orwellian future that awaits. For others who have been following this space, the outcome appears to be a natural evolution of data science and psychological science. Regardless of your position here, one thing is clear: the future seems both terrifying and exciting.
Ethical breaches within data governance and threats to the democratic process aside, perhaps the most interesting take-away, for me, is the complete realisation of the power of psychology when combined and leveraged with technology. Despite being over a century old as a profession, the applied practice of psychology has been somewhat non-mainstream – like a supporting actor in a movie or a drummer in a band (sincerest apologies for any drummers reading this). Yet with the wake of the Cambridge Analytica scandal, psychology, in particular the field of psychometrics and profiling, has taken centre stage by influencing the election results of one of the world’s most powerful nations. With this grand entrance to the mainstream media, psychology ironically finds itself both in the spotlight and under the microscope.
How do we reconcile the fact that the combination of psychological science, technology, and data science can understand us better than ourselves? What are some other use cases of psychological profiling today, and what are some of the considerations that exist? In this article, I attempt to discuss my reflections on the topic and invite you to join me on this process of discovery.
A Brief History of Modern Psychological Profiling
When one thinks of psychological profiling, one typically thinks criminal profiling – where television shows like Criminal Minds have portrayed the method as one whereby a set of characteristics are strung together or deduced into a profile which is then used to help solve criminal cases. Forensic psychology and criminal profiling, while interesting in its own right is not the direction that I want this article to follow. Instead, I’m going to focus on the application of psychological profiling within the business world, or more specifically within talent management and marketing/user experience where the basic premise is still the same: build a model or an abstraction of a person or type of person based on a combination of psychometric measures such as personality and IQ/cognitive ability and make predictions and tailor strategies around those profiles.
Within the personnel or HR space, psychological profiling began in World War 1 by the U.S., ARMY to identify future officers via the use of intelligence and personality assessments. Typically, a profile within here is built around the idea of an ideal person for the job, which takes into account the job requirements, the external and political environment, and filters down into the characteristics that the job incumbent should have. Using the example of an ARMY officer, qualities such as intelligence, assertiveness, resilience may form that profile, and psychometric methods are then used to measure those qualities. Building on nearly 100 years of research, this branch of profiling (that is, the ideal person for the job), resides within the domain of industrial-organisational psychology with companies like Saville Assessments and CEB leading the way with assessment tools for cognitive ability and Hogan Assessments for personality. Today, up to 80% of the fortune 500 companies use some form of psychometric assessment for purposes of profiling for the selection and development of employees.
Within the marketing and consumer decision-making domain, the use of psychological profiling has been much more sporadic, owing to the fact that consumer psychology is still largely an academic discipline and perhaps a lack of engagement by consumer psychologists. Still, profiling has seen some use within market research firms and advertising agencies from as early as the 70’s. In contrast to profiling within talent management which focuses on understanding how people might fit in or perform within their jobs, the objective here is to put together a picture of the buyer or the end user of a product or service. Today, we see an adoption of user personas within the UX and consumer research space which draws on personality and individual differences to inform design choices and continued, albeit low levels of adoption within the advertising space. This is evident in advertising firms that claim to drive behaviour change through the application of psychological theory – companies like Naked Communications (previously headed by Australian consumer psychologist Adam Ferrier) and of course, Cambridge Analytica. The bottom line here is that the consumer and marketing side of profiling has been lagging behind, but that tide is changing because:
As we enter the technological revolution that is industry 4.0, the idea of understanding people – whether it be the user or the employee is important now more than ever.
The Current Landscape
The rise of Cambridge Analytica represents a leveling of the playing field for psychological profiling as the consumer profiling side of the industry catches up to the employee profiling both on a technical level (i.e., large scale data science to measure and predict) as well as a commercial level by being one of the only firms to offer end to end research and marketing services based on psychological science. Though mired in alleged questionable business practices, Cambridge Analytica still stands as a pioneer within this space by successfully combining data science, technology, and psychological profiling to create a prediction engine that is powerful enough to supposedly influence the Brexit campaign, as well as the U.S., elections. Cambridge Analytica’s fate is unknown at the time of writing this, but the genie has been let out and it is now only a matter of time before others like it – companies that combine data science and technology with behavioural science- will rise from the ashes. Will we see another end-to-end company that integrates psychological research into every process? Perhaps. As the science of psychological profiling matures and technology and data science improves, other use-cases (perhaps not so politically charged) will naturally emerge. In the sections that follow, I discuss two that I see within my horizon.
One of the lesser known areas of talent management is the interplay between the physical environment and employee wellbeing and productivity. Indeed, talent management is often thought as a process of finding the best talent for the job, and ensuring that there are proper organisational processes and resources to ensure that they can perform. What is often missing is an understanding of where and when the work happens or the role of the physical work-space. Examples of this are seen within the open plan trend where open designs are implemented without much thought into how it affects the individual or groups. Advocates of open plan offices propose that in addition to reducing costs, open plan offices improve communication and by extension, collaboration. The reality is that the research on this is actually mixed – open plan offices, in some case, can lead to a decrease in performance (depending on the type of work), decrease collaboration, and even worsen levels of mental health. At Rate my Space (a tech startup that helps organisations and property managers create user insights for buildings, workplaces and campuses), we found that people within open plan configurations reported lower levels of two key aspects of mental wellbeing – achievement and meaning. The finding, though intuitive, runs counter to some of the claims that are often attributed to open plan work spaces because:
For many of us, work is the place where we derive much of our meaning and sense of accomplishment from. The finding here is intuitive – take something away from people without giving back something commensurate and people will feel worse, and likely perform worse.
A skew or unchecked preference towards open plan offices also ignores the fact that people, with their unique personality and preferences, work better in different environments. A completely open plan environment might be beneficial for a sales team (typically on the higher end of the extroversion personality spectrum), whereas an engineering team (typically on the lower end of that spectrum) will find themselves less productive. This is clearly an oversimplification and not an attack on open plan offices per se, but a call to action to integrate psychological science into the design of work and of the physical workspace.
The function of a workplace has evolved from simply being a place to conduct business to a place either promotes or diminishes physical and mental health, a place that inspires creativity and innovation or promotes anxiety and depression. With pioneering research from Psychology and related disciplines, we now have a better understanding of the influence of the work environment – both physical and social on productivity and wellbeing. We know, for example, about the regenerative properties of a physical workspace (i.e., the influence of office plants/greenery on stress reduction) and the type of environment and temperaments that are optimal for specific tasks (e.g., the type of environment and its relation to creative output and quiet spaces for “deep work”). These work preferences are also influenced by psychological factors such as personality and cognitive and emotional states, which is to say that we may prefer engage with different office amenities differently according to personality and how we think and feel.
Collecting all this data and applying the magic sauce that is data science with emerging technologies seems to be a perfect marriage. For example, in the sensor driven world that we are approaching, the IoT and more specifically, technology enabled workplaces allow for more data than we can possibly want. We have sensors that can monitor things like indoor air quality and temperature, movement and posture to emotions and mood with wearable technologies that measure galvanic skin response, heart rate, and skin temperature (all known correlates or markers of mood). This trend towards “emotion sensing” technology represents an exciting direction when combined with psychological data because it opens up the opportunity to a truly smart workplace.
What if instead of providing a static workplace, which is effectively a “catch all” solution, we provided an adaptive building or floor that physically changes based on who is there, how many people occupy the space, and the current activities that are planned for the day? A workspace that provides people with the control to influence the environment, but also the intelligence to modify things like lightning, mood, presence of greenery, noise suppression based on individual differences in personality and preferences. A workplace that physical transforms and adapts to close off physical segments for deep work and to open up for collaborative work or based on the ratio of different psychological profiles of people that occupy the building at that time. While we are a few years away from something like this, it nonetheless is an exciting prospect, and one that is enabled by a deep understanding of user behaviour vis-à-vistheir psychological profiles. With the understanding that the workplace is a fundamental source for wellbeing, on top of a place to innovate and conduct business, organisations now are under more pressure to ensure that the physical workspace, among other things, can truly enable those outcomes. It’s time we stopped ignoring the role of the physical environment.
From chatbots to autonomous cars to working androids, the future of work is one where Robots and AI are inseparable from humans. While the distant end-game of automation and robotics may be one where robots will eventually replace many existing jobs (and then the dominion of man-kind in general?), the problem that people and organisations will face in the near-term is how to work alongside these systems effectively. Many autonomous systems today are designed with human limitations in mind – things like attention, memory, and information processing are all carefully implemented into design to ensure that the user is positioned to take action when they are required to. As autonomous systems and robots become smarter and more self sufficient (i.e., operating without any human intervention), and as we move beyond simply interfacing with autonomous systems, the next frontier lies within designing robots that can work effectively with people – robots that are trusted and cooperative. This is especially important as robots and autonomous systems become more commonplace and “integrated” within teams because the productive output of a team is contingent on how well everyone works together. Here, psychologically profiling appears to be the logical next step – we already design systems catering to cognitive limitations, adding personality and user preference into the mix seems like a natural evolution. In the same way that Cambridge Analytica breaks down people into categories based on their personality (e.g., low conscientiousness, high neurotic, low openness to experience) and tailors specific messages to that segment, the same could be applied to working robots. Rather than having a single universal speech response or action, robots instead behave and speak in ways that is consistent to the expectations of the end user or complimentary to the team they’re situated in.
Sometimes the situation requires for a stoic and conscientious R2D2, others, an extroverted and “creative” C3p0.
This technology currently exists, powered by a natural language processing (a core area of artificial intelligence) and we already see some of this work within the design of chatbots. Startups like X.Ai – a chatbot that intelligently schedules meetings via emails is powered by a team of writers and people with creative backgrounds. By approximating the “Big Five” traits (the same one used by Cambridge Analytica), the creators personified their bots as Amy and Andrew – prototypes of the ideal personal assistant based on a personality type. Something interesting to note is that with a chatbot, the interface is simplistic and therefore the UX is entirely dependant on personality, meaning that the “product” isnt defined by its physical properties but rather how likeable and human-like it is. I can imagine that the next step here will be to throw fuel onto the fire by adding data science into the mix to model different human personality types with response from each robot persona – effectively training the chatbots to adapt and mirror the end user to create a sense of trust. This is personality by design, and it’s the way of the future.
Moving away from the chatbot space and into actual robots, researchers from MIT are investigating the influence of a robot’s behaviour on human colleagues within the workplace. By mimicking human behaviours such as leaning forward, verbally expressing, listening, and even showing initiative (i.e., moving and helping before being told to), these robots were perceived as having more commitment to the team and instilled a sense of confidence and trust within the human team members. These subtle human gestures form the basis for empathy (a state that is often taken for granted by humans!) and is crucial for any collaborative effort. Indeed, as we move beyond querying Siri or Alexa for a response to a true collaboration like a football team – effectively working as “one”, the idea of a psychological profile for a machine will sound less science fiction and more commonplace. Here, we move on from machines that are designed to optimally engage with humans to the idea of a fully fledged AI that come with its own personality, or dynamic way of engaging with humans. One of the necessary preconditions for high performing teams is presence of psychological safety, an underlying sense of trust within the team and its members. What happens when we add a robot into the mix? Do we pre-load these robots with the personality profiles of each team member and have it programmed to respond optimally in that way? Is it even the robot’s responsibility? – in other words, if we replace a human within that team, does the emotional labour that comes with managing the cognitive and emotional states within the team get passed onto the existing human members? And if this is the case, what are the trade-offs that need to be considered? Unfortunately the answers here lie within the realm of speculation, but they’re going to need to be answered as we approach the near future.
In both these cases, we see a move away from a shallow engagement with machines to a process of collaboration and co-creation. Bots and robots are now going through a process of anthropomorphism with specific design choices to become more human-like and with that, brings implications for how we engage with these intelligent systems. Whether it be designing chatbots that generate a great user-experience to robots employees that people trust, the role importance of psychology and personality becomes more prominent. In the same way that the physical work environment is now an enabler for human endeavours such as creativity, innovation, and wellbeing, human-robot interactions within the workplace contribute to the human experience – how we work, how happy we are, how engaged we are will be influenced and mediated by our interaction with these intelligent systems.
Throughout history, humans have often engaged others with a blank slate – a tabula rasa of sorts, and a reasonable degree of trust. With prolonged engagements, we then build a picture of that individual – their tendencies, proclivities, and preferences. With more time, we can say with some confidence that we know what their personality may be and eventually get to a place where we “know” someone. That process of knowing – whether it by with your own self or of others, serves as a powerful mechanism for interacting with others and is -dare I say- what life is all about. Organisations today hold immense power by having access to psychological profiles of their employees and of their consumers through powerful psychometric assessments. To drop one of the cheesiest quotes that exists: with great power comes great responsibility. When it comes to influencing other people, there exists a thin line between nudging behaviours to outright manipulation. By building technologies that can comprehend human traits such as personality and emotions and to be able to act in ways that influence behaviour, we improve the user experience and build better products and services, but the tradeoff is that we run the risk of compromising free will and ultimately the foundations of what it means to be human. To say that it is a minefield is an understatement, but alas, there is a solution.
Organisations need to have a chief psychological officer. The title of a chief psychological officer currently exists within psychological consultancies, typically within assessment and their core function is to lead the assessment or consulting divisions. I’m proposing that this role needs to be created for any company that combines data with psychological science, whether it be for profiling or other uses. Two main reasons stand out for me: subject mater expertise and an ethical framework. Drawing on a minimum of 6 years full time training, psychologists are subject matter experts within domains such as personality, mental health, and much more. With current industry demands such as the focus on the user experience and employee experience, it stands to reason that psychologists – with their background in the science and practice of psychology would be well positioned for the job. We also see the lines blurring between marketing and HR, so having a chief psychological officer would help to bridge the gap, to break away from the silos because the science of understanding people is the same for the customer and the employee. They’re also data literate (though some have a higher tolerance than others), which fits in nicely with the big data and analytics trend that is capturing global organisations.
Psychologists are also ethically bound to a code of conduct that permits or mitigates harm to the general public – a covenant that is tied to their licence to practice- which is especially important for the moral and ethical ground that applying psychological tends to be in. Would Cambridge Analytica have been better positioned with a chief psychological officer? It’s hard to say with any degree of certainty without understanding the internal politics, and it’s a job that retrospectively, nobody would want right now! My point here is that the chief psychological officer could be a bulwark against malevolent applications of psychological methods, a first line of defence. At the very least, a psychologist would be obligated to steer the company towards minimum harm if there was a perceived risk. Urging oranisations to understand what psychologists bring is not enough. The other side of the coin lies with psychologists. We need to extend beyond the therapy couch or consulting room to learn the language of technology – to engage with industry and be involved in shaping the future. The stakes are too high.