Recent years have seen an upsurge of concern in the technical and popular press about the impact of automation upon employment in western societies. Once the concerns about the development of artificial intelligence were concentrated within the realms of computer science and philosophy, debating the extent to which a robot could be conceived to have self-awareness and consciousness. Those concerns have migrated to the social and business pages where the increasing advances of automation of jobs and skills are being recognised. The trend has been to point to the downsides of automation and the rise of artificial intelligence. This has moved further from a concern about how lower skilled jobs, such as in mass production of cars and white goods, can be supplemented by robots to a perception that the skills of many professions are in line for replacement. This leads in turn to contemplating the impact of automation on the profession of psychology.
This is a virtual interviewer created by scientists at the Institute for Creative Technologies at the University of Southern California.
Undoubtedly, the current impact of technology on the profession of psychology has been largely positive. Advances in online communication have paved the way to developments in online therapies, which have bolstered opportunities for accessing psychological services; typically a major challenge for those living in remote areas. Additionally, the ‘app’ revolution is affecting the delivery of psychological services, augmenting the professional practice ‘toolkit’, with a range of user-friendly and convenient programs. These may enhance practitioners’ treatment approaches, improve the public’s awareness of mental health issues, and provide clients with fingertip access to professional support. A growing number of apps even work to equip the user with evidence-based treatment techniques (e.g., cognitive behavioural therapy), which enable them to identify and change unhelpful thinking styles by themselves (e.g., MindShift).
Concerns have been raised about the quality of these programs and their effectiveness, and as such, they may not be as strong a threat to psychological practice as first appeared. However, IBM predicts that in the next five years, advances in cognitive robotics will enable machines to analyse patterns in our speech and writing to detect early-stage developmental disorders, mental illness and degenerative neurological diseases (Cecchi, 2016). The speed and quality of technological development is likely to overcome such doubts. The outsourcing of psychological tasks to automation has been occurring for some time (e.g., assessment scoring, report writing) and demonstrates the profession’s willingness to embrace the rapidly evolving technological landscape. However, it also raises critical questions for psychologists of the 21st century: What are the long-term impacts of advancements in technology to our profession? How much of what we do can (and should) be automated? Will the introduction of automation and robots reduce the role of the psychologist, and ultimately, the demand for human psychologists?
Automation of psychology skills
Psychology is seen as a profession which requires the application of interpersonal skills and subtle abilities of formulation and interpretation and empathic understanding that are highly unlikely, if ever, to be achievable by machines and so unlikely to be replaced by them. However, it could be argued that more and more of the so-called soft skills of the psychologist (e.g., active listening, positive regard, the reading of non-verbal cues) are being captured by automated processes (see Ellie the virtual therapist; Rizzo et al., 2015), which could result in the wholesale replacement of psychologists. Even if it can be shown that there are elements of the job of a psychologist that may never be replaceable, the process of automation will nevertheless result in many components of the psychologist’s job being replaced, which will result in a need for fewer psychologists than are presently required and who are being trained.
Psychology as a profession, therefore, may not be exempt from the advance of the robot. The argument hinges on a feature of the development of thinking about artificial intelligence over the past decades. There is a ‘strong’ version in which artificial intelligence can be created that demonstrates not only the power to reason, remember, plan, learn and communicate, but can also show consciousness, self-awareness and feelings. Portrayals in the media have overwhelmingly focused on such features. The computer HAL 9000 in Stanley Kubrick’s 2001: A Space Odyssey (1968), the achievement of ‘self-awareness’ by Skynet on 29 August 1997 in James Cameron’s Terminator, and the female robot, Ava, in Alex Garland’s Ex Machina, all demonstrate machines with feelings and consciousness that are a threat to the existence of mankind. The warnings of scientists such as Stephen Hawking about the threats to our existence as a species (Hawking, 2016) are based largely on an understanding of the ability of robots to think and feel like humans, only more logically and with less responsiveness to emotion. However, the possible threat to the employment of humans as psychologists does not rely on the attainment of such high-level skills. The replacement of humans may be achievable by much less esoteric mechanisms.
So far as we are aware, the strong version of artificial intelligence has not yet been achieved, and it remains unknown whether it will ever be possible to do so. Weaker versions of artificial intelligence are constantly being developed and are available today, doing jobs that once were seen as wholly in the province of humans. Highly constrained versions of intelligence are being developed in particular contexts. The development of very fast computers and large databases make it possible for computers to access data and respond to questions rapidly and reliably, and simulate human decision-making. It is not necessary to replicate the ways in which humans think and make decisions as algorithms exist that can simulate the actions and provide answers (Susskind & Susskind, 2015). The defeat of the chess grandmaster Gary Kasparov by the IBM computer Deep Blue showed the developing power of automation, and the March 2016 defeat of the world Go champion, Lee Sedol, by an artificial-intelligence program developed by Google was the final victory of computers over human game players.
Predicting future automation
With the realisation that weaker versions of artificial intelligence could perform operations which mimicked human decision-making it has been possible to set out what might be regarded as ‘bottlenecks’ to the intrusion of automation into human activities. Frey and Osborne (2013) identified three sources of such barriers as involving perception and manipulation, creative intelligence and social intelligence. The authors in this landmark study made an initial subjective assessment of the requirements of 702 jobs. Looking at a subset of these jobs, the authors assessed the degree to which “the tasks of the job are sufficiently specified, conditional on the availability of big data, to be performed by state-of-the-art computer-controlled equipment” (Frey & Osborne, 2013, p.30). It is important to note several aspects of this process. The authors refer to the availability of big data and state-of-the-art computing. This implies that with the rapid development of both of these engineering concepts there will be rapid expansion of the range of jobs, or at least the sub-components of these jobs, which can be replaced by computers. It is crucial to recognise that the authors made a subjective assessment of what components underlay the jobs in order to write the algorithms to generate their predictions. Ignorance of what is involved in any particular job could result in a serious underestimation of the extent to which the job could be replaced.
Ava, from the movie Ex Machina with a robotic body but a human-looking face is depicted as convincingly passing as a human.
Frey and Osborne (2013) were then able to list the jobs in rank order of the probability with which they were likely to be replaced in the years to come. The likelihoods included 98 per cent for the replacement of umpires and referees, 96 per cent for cooks, 64 per cent for stock clerks, 37 per cent for actors and 4.7 per cent for mathematicians. Debate may be likely in the assessment of the estimates for many jobs on the list. The prospect of actors on stage and cinema being replaced does seem low. However, even here the advent of computer-generated imaging in cinema does indicate a higher likelihood of human actors being replaced, as evidenced by the digital representation of the long-dead actor Peter Cushing in the Star Wars vehicle, Rogue One.
With only a likelihood of 0.43 per cent of being replaced, the job of the psychologist, seen by Frey and Osborne (2013) as heavily dependent on the application of skills involving social intelligence, skills of negotiation, persuasion and the caring for others, should continue to provide job opportunities for decades to come. However, is this estimate realistic or is it a failure to appreciate what a psychologist does and what the job requires? Even if only parts of the job of a psychologist are replaced, what jobs will remain and how many people will be required?
Presently, psychology has fundamentally embraced the fruits of technology. However, consistent with this open-mindedness to change, we should continue to look ahead to what increasingly accelerating technology may bring, appreciate those continuing developments and how they may change the face of the profession in a very short period of time. For now, the face of psychology remains largely a human one. But tomorrow?
The first author can be contacted at [email protected].