CHAPTER 6

Thinking in 4T: Helping Workers Help Themselves With AI

Terri Griffith

Abstract

You see in 3D, and the future of work requires that all of us, not just organizational leaders, think (and act) in 4T. We all need fluency in how to integrate all of our talents, technology, and techniques as we work toward the targets of whatever our work is—whether paid work or the work of our lives. No silver bullets (Brooks 1987): relying on the dimension we’re most comfortable with or have at hand.

This chapter is motivated by the growing application of artificial intelligence (AI) to work at all levels. For this chapter, I consider AI as a broad class of tools including hardware and software robots and intelligent devices. From basic rule-based automation to AI built on strategies of deep learning, we can apply new tools across much of our work. Possibilities range across colorizing movies, fraud detection, marketing lead generation, trip planning based on traffic conditions, foreign language translation, and depositing checks.

Top-Down and Bottom-Up

The application of AI and more basic automation is a top-down and bottom-up effort. Certainly, organizational management plays a massive role in providing direction and resources for next steps within the organization. The business press is rich with well-presented books on how to integrate automation into organizations. This is the most extensive and complete coverage I’ve ever seen on a topic I follow. A few of the excellent examples published this year:

• Prediction Machines: The Simple Economics of Artificial Intelligence (Agrawal, Gans and Goldfarb 2018)
• Human + Machine: Reimagining Work in the Age of AI (Daugherty and Wilson 2018)
• Reinventing Jobs: A 4-Step Approach for Applying Automation to Work (Jesuthasan and Boudreau 2018)

Each of these books offers a fairly top-down approach where the focus is on management action. Each does touch on, albeit briefly, what individual workers and teams should be doing to take advantage of resources and practices available today and how to prepare for the future. I extend from these AI/automation books in how individuals can, and should, work from the bottom-up.

The bottom-up approach to automation requires a different way of thinking and a different allocation of responsibility for job design than we typically see in the application of technology tools or management practices. Thinking in 4T requires evaluation and orchestrated application of talent, technology, and technique in service of specific targets. These ideas are aligned with academic studies of sociotechnical approaches to work (Trist and Bamforth 1951), and more recently, sociomaterial approaches (Cecez-Kecmanovic et al. 2014). Unfortunately, while broadly applicable, neither approach has received the kind of widespread attention and use that we see with more deterministic tools like Six Sigma approaches to process improvement or agile methods in software development.

Smart, well-resourced individuals and organizations predict vastly different overall effects on jobs as we see greater introductions of artificial intelligence and other forms of automation (Agrawal, Gans and Goldfarb 2019). The one thing experts agree on is that tasks within work will change and that this change will affect some workers more than others. Thinking in 4T is one way to prepare for these changes. It may be that individuals who Think in 4T as they adapt their work will be more protected from future job pressures than individuals who let the organization do the adapting from the top-down.

Why Contrast 3D and 4T?

Because 68 years of calling these ideas “sociotechnical systems theory” (Trist and Bamforth 1951) hasn’t moved this critical idea into the mainstream. Work by Nobel Laureate Daniel Kahneman (2011) and others highlight how hard it can be to think with agency around complex issues. We need signposts to kickstart this thinking, and Thinking in 4T can be the first step. Marketing research suggests that simple tropes, like the juxtaposition of 3D and 4T, may serve as good reminders (Toncar and Munch 2003).

Connecting the Ts

Decades of research, starting with sociotechnical systems theory (Trist and Bamforth 1951) as noted earlier, have championed a systems approach to organization and work design (Parker, Morgeson and Johns 2017). Perhaps the most famous in technology settings is the imagery of ­Frederick Brooks, an A.M. Turing Award winner and team leader on the IBM System/360. Brooks writes, “...we see no silver bullet. There is no single development, in either technology or in management technique, that by itself promises even one order of magnitude improvement in productivity, in reliability, in simplicity” (1987, p. 10). We need to manage the 4Ts in concert.

Helping Workers Help Themselves With AI

Awareness of Artificial Intelligence

While Thinking in 4T was important in the past, it is now critical at all levels of organization and life. Institutional and individual interest in AI is on the rise. The National Science Foundation seems aligned with this perspective, citing “The Future of Work at the Human-Technology Frontier” (2018) as one of its 10 big ideas. Their focus: “Understanding how constantly evolving technologies are actively shaping the lives of workers and how people, in turn, can shape those technologies, especially in the world of work.”

Google searches for branded versions of automation are up dramatically since November 1, 2014 using the month Amazon Echo launched for the baseline (see Figure 6.1). However, even as we become more aware of the concept of AI, the AI technology lacks many of the triggers (e.g., Griffith 1999) that could help us think effectively about how to integrate it into our work. Even when we are aware that we are using AI, many of AI’s functions can be a black box (Verghese, Shah and Harrington 2018).

Figure 6.1 Worldwide relative search frequency for two branded automation platforms

Value of Thinking in 4T

Prior research suggests that the bottom-up approach to leveraging AI at work will be valuable to individual workers. Brenninkmeijer and Hekkert-Koning (2015) find that people who craft their work can increase their performance and employability. A worker who takes on the individual challenge of understanding his or her target, talent (knowledge, skills, and abilities), available technologies, and the techniques to bring these dimensions together within their role will be more likely to race with, rather than against, the “machines” (Brynjolfsson and McAfee 2011). That is, rather than go on the fool’s errand of trying to best an AI or other automation on, for example, well-structured tasks that don’t require dexterity, physical skill, or mobility, Brynjolfsson and Mitchell (2017) suggest using AI to augment tasks that leverage our human creativity and interpersonal skills.

It may also be that 4T-thinking workers find better roles as organizations transform within the Fourth Industrial Revolution (Schwab 2017). Thinking in 4T and taking action based on that thinking is critical for all of us. We all know our own work best. Thinking in 4T is a chance to take that work to the next level in a world of increasing change. Quoting Sophocles, “...heaven ne’er helps the men who will not act” (Plumptre 1978, p. 406).

That said, in organizational settings, work design that is both ­top-down and bottom-up (negotiated) has the most robust outcomes (Hornung
et al. 2010). Let’s help workers help themselves.

Triggering 4T Thinking

Applied psychology helps to explain why the application of integrated approaches like Thinking in 4T are difficult. Kahneman’s (2011) best-selling general audience book, Thinking, Fast and Slow, highlights work on System 1 and System 2 modes of thinking, and this is an effective approach to understand the barriers and opportunities for augmenting work with automation. System 1 is automatic, quick, requires little or no effort, and presents no sense of voluntary control. System 2 thinking requires concentration, is complex, and has agency and choice. Kahneman offers an opening example (p. 19) of System 1 thinking in the form of a picture of an angry woman with tense eyes and an open mouth. He predicts, correctly in my case, that the reader’s understanding of anger and likely next actions (e.g., speaking loud, unkind words) came instantly. His second example is the multiplication problem 17 × 24. Knowledge that is a multiplication problem came instantly (part of System 1), but the process to reach the answer would require steps of deciding whether or not to take on the task and then developing a process to do so (e.g., planning to hold the numbers in my head or getting a pencil and paper). This more procedural work is an example of System 2.

I see extant practices around change at work as being closer to System 1 than System 2. When we implement a change without clearly identifying our target, talent, technology, and technique we are more likely to go with what we know. We apply the status quo bias (Samuelson and Zeckhauser 1988). In the most static form, we opt to continue with how we are already doing our work. In a slightly more dynamic version, the easier we can substitute AI for something we already do, the better rather than considering whether the overall approach should be adjusted.

Thinking in 4T requires a System 2 mindset. Thinking in 4T requires that we stop and consider our goals and available resources, consider trade-offs across the different resources, and then a practice to put them into play relevant to our work. System 2, Thinking in 4T in this case, is a challenge “if you are not ready or if your attention is directed inappropriately” (Kahneman 2011, p. 22).

Nudge Perspective

I’ll offer six strategies for triggering Thinking in 4T. These build from another best-selling general audience book: Nudge by Nobel Laureate Richard Thaler and Cass Sunstein (2008). Thaler and Sunstein admit to taking some liberties with the initials, but they come up with a mnemonic (itself a nudge) to present their application of a variety of economic and psychological supports for changing behavior. Here I provide their nudge descriptor and then research supporting the approach as we consider Thinking in 4T and helping workers help themselves by augmenting their work with AI. The goal at the highest level is to create situations where using the 4T approach is easy and avoiding its use is more difficult.

Incentives

Intrinsic and extrinsic incentives are both predictors of performance, with intrinsic motivation playing a stronger role when there are less direct connections between performance and reward (Cerasoli, Nicklin and Ford 2014). Both intrinsic and extrinsic incentives can trigger the use of Thinking in 4T, or any other approach around the augmentation of work via AI or other automation. To the extent that workers see value outweighing the costs (or are helped to see), they are more inclined to act. Whether designing work for ourselves or for others, we need to highlight the connection between implementing AI in an integrative way and highlighting the opportunity for rewards.

Understand mappings. Situations where experimenters ask participants to do something, even something uncomfortable (e.g., get a tetanus shot), and then have participants specifically map how they will do it engender greater compliance than settings with general admonishments. (Leventhal, Singer and Jones 1965). I propose that Thinking in 4T to apply AI to an individual’s work is much more enjoyable than a tetanus shot, but nonetheless we should apply similar techniques. Following from Gollwitzer (1999), helping workers create memorable if-then statements along the lines of, when thinking about how to take on a new task, I will take a stop-look-listen (Griffith 2012) moment, evaluate my 4T options, and then design an approach that takes each of the 4Ts into consideration. It may be that the best way to approach the target does not involve equal parts of talent, technology, or technique. It may be that the best application will use a subset of the Ts, though their interactions should be considered in order to find the best approach for the setting.

Defaults. The defaults nudge offers two distinct approaches: set the default such that AI/automation/augmentation of work is the default or acknowledge the status quo bias and put in place triggers for active thinking. Louis and Sutton (1991) offer three approaches for switching from automatic (System 1) to active thinking (System 2): novelty (a situation that is perceived as new), discrepancy (a situation that is perceived as different from that expected), and deliberate requests for active thinking (a situation that may parallel the mapping request in the prior nudge). Griffith (1999) applied this perspective to evaluate triggers for sensemaking around new technologies and notes that it is possible to design these triggers into the technology itself. AI or automation features that are specifically called out and highlighted as more core (versus tangential) and more concrete (versus abstract) will more likely trigger sensemaking around effective use, breaking people free from default approaches.

Give Feedback

It is hard to make good choices if we can’t learn from choices made in the past (and ideally, also what would have happened had other choices been made). Task-provided feedback, where outcomes are directly observed during performance of the task, is the most immediate (Hall and Lawler 1968) and accurate source of feedback (Campbell et al. 1970). Task feedback is more informative than feedback originating from the supervisor (Greller and Herold 1975).

Expect error. In the physical world, this nudge is more of a preemptive guardrail. Thaler and Sunstein (2008) offer a variety of examples. Gas-specific connectors in medical environments such that misconnections are physically impossible. Google’s backend automation within e-mail: When you attempt to send an e-mail that contains the word “attachment,” but you have not attached anything, the system prompts you before allowing the send. Medication regimes being built to allow for more likely compliance—drugs that can be taken once a day in the morning, drugs packaged so that a pill is taken daily, even though the active medication is only 21 days a month. Designers understand that people fall back on habits when they can (i.e., not blocked by a connector that doesn’t fit or an e-mail that doesn’t send), or when context doesn’t offer an incentive to avoid the error (e.g., Wood 2017, Louis and Sutton 1991). Cars with autonomous emergency braking offer a clear example of the value of this nudge. Use of this technology is documented to offer a 38 percent reduction in rear-end accidents (Fildes et al. 2015).

Structure Complex Choices

This last nudge is exactly what this chapter is about. For workers to integrate technology into their practice requires a complex negotiation across targets, talent, technology, and technique. To the extent that we can help people by offering a structure for their complex choices (focused on 4Ts, think of the mixing across the dimensions as a negotiation, Griffith 2012), we should.

Managing and Measuring Thinking in 4T

Once we trigger Thinking in 4T, we need to manage the approach like any other change. A focus on ability, motivation, and opportunity can drive toward performance (Blumberg and Pringle 1982). We can apply some Thinking in 4T ourselves. Our target is a shift in mindset from easier, System 1, silver bullet approaches to a goal-driven, multidimensional approach. The time seems right for this. Salas, Kozlowski, and Chen (2017), for example, offer the following in their 100-year review of applied psychology, “We need to start thinking in terms of human-system integration (e.g., Kozlowski et al. 2015) and think how selection, training, socialization, mentoring, leadership, and other domains will look like in the next century when technology is embedded in most organizational systems” (p. 595).

Talent comes in many forms. Earlier, I suggested some individual-level strategies for triggering Thinking in 4T. Much of our work is in teams and the growth of virtual teams in organizations is explosive (Dulebohn and Hoch 2017). While individual characteristics (knowledge, attitudes, traits, and abilities) are foundational to team practices and outcomes, teamwork, and innovation is an emergent, relational process (e.g., Burke et al. 2006). Grant (2009) notes:

Relational perspectives focus on how jobs, roles, and tasks are more socially embedded than ever before, based on increases in interdependence and interactions with coworkers and service recipients. Proactive perspectives capture the growing importance of employees taking initiative to anticipate and create changes in how work is performed, based on increases in uncertainty and dynamism (p. 317)

We need to develop team-focused introductions to Thinking in 4T. Virtual teams, especially, use technology. Thinking in 4T for teams, and specifically, the application of AI, is powerful in such a setting.

Measurement is critical as we track across target, talent, technology, and technique. Goodman and Griffith (1991) offer a layered strategy that can be applied to measuring implementation success:

• Knowledge
• Attitudes
• Behavior/performance
• Normative consensus

The layered strategy acknowledges the dynamic nature of implementing any change. In the context of Thinking in 4T, the first layer is knowledge of the approach. Chapters/articles/blog posts, workshops, course sessions, and webinars are all part of the process I am using to share the ideas of Thinking in 4T. Measurement of knowledge is around knowledge of the four Ts, the ability to explain the use and value to others, and steps to using in individual, team, or organizational levels.

Attitudes come next as they, for better or worse, can precede interaction with the new tool or approach. Positive or negative feelings may ebb and flow as knowledge and experience buildup, and continued measurement can help with managing the process.

Behavior/performance can be mental or physical: Are people planning their application of AI to their work with a 4T approach? Is the identification of work targets part of the process and are resources considered across each of the talent, technology, and technique dimensions?

Normative consensus supports the focus on the relational aspects of the process. Once an application of AI is in place, is there a sharing process (Griffith 2012) that offers opportunities for reflection, improvement, and agreement that work should head in a particular direction. How widely held is the assessment? Does the approach diffuse beyond the initial application?

Concluding Remarks

My use of two best-selling books (Nudge; Thinking, Fast and Slow) is strategic. Thaler, Sunstein, and Kahneman are renowned scholars who found ways to describe evidence-based approaches to work and life that resonate with the broad audience we need to reach. Leveraging their effective language may offer deep support for the Thinking in 4T brand. Binding Thinking in 4T to the idea of seeing in 3D offers a trigger (Louis and Sutton 1991; Griffith 1999) to disrupt System 1 approaches to work and life with constantly changing technologies.

Responsibility for sharing new ways of thinking and working falls to all of us who understand that a silver bullet approach to how we engage with AI and other automation can’t succeed. We should hold workshops on Thinking in 4T, design thinking, systems thinking, and so on. Lunch and Learns with vendors of light-weight AI (even if only around tools like Apple’s Shortcuts app) may trigger colleagues to think about how they can individually craft their work. Face-to-face or virtual communities of practice (e.g., Griffith and Sawyer 2009) offer the chance to share success, get feedback, and develop norms around new ways of working.

And finally, we are in positions to continue to build evidence-based approaches to shifting new ways of working. Collaborations across professional networks, scholars, vendors of tools, and other organizations offer opportunities to share the ideas and the value of Thinking in 4T. Future research should first test these applications of psychology in work settings where AI offers clear value. Follow-on research could test different training tools, and evaluate how early in a person’s experience the perspective provides value. We can assess technology designs in terms of their ability to trigger thinking in 4T (e.g., Griffith, 1991) or the value of default settings (e.g., Thaler and Sunstein 2009) that maintain System 2 modes.

References

Agrawal, A.K., J. Gans, and A. Goldfarb, eds. 2019. The Economics of Artificial Intelligence: An Agenda. Chicago, IL: University of Chicago Press.

Agrawal, A., J. Gans, and A. Goldfarb. 2018. Prediction Machines: The Simple Economics of Artificial Intelligence. Cambridge, MA: Harvard Business Press.

Blumberg, M., and C.D. Pringle. 1982. “The Missing Opportunity in Organizational Research: Some Implications for a Theory of Work Performance.” Academy of Management Review 7, no. 4, pp. 560–569.

Brenninkmeijer, V., and M. Hekkert-Koning. 2015. “To Craft or Not to Craft: The Relationships between Regulatory Focus, Job Crafting and Work Outcomes.” Career Development International 20, no. 2, pp. 147–162.

Brooks, F. 1987. “No Silver Bullet—Essence and Accident in Software Engineering.” IEEE Computer 20, no. 4, pp. 10–19.

Brynjolfsson, E., and A. McAfee. 2011. Race against the Machine: How the Digital Revolution Is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy. Lexington, MA: Digital Frontier Press.

Brynjolfsson, E., and T. Mitchell. 2017. “What Can Machine Learning Do? Workforce Implications.” Science 358, no. 6370, pp. 1530–1534.

Burke, C.S., K.C. Stagl, E. Salas, L. Pierce, and D. Kendall. 2006. “Understanding Team Adaptation: A Conceptual Analysis and Model.” Journal of Applied Psychology 91, no. 6, pp. 1189–1207.

Campbell, J.P., E. Dunnette, E. Lawler, and K.E. Weick. 1970. Managerial Behavior, Performance and Effectiveness. New York: McGraw-Hill.

Cecez-Kecmanovic, D., R.D. Galliers, O. Henfridsson, S. Newell, and R. Vidgen. 2014. “The Sociomateriality of Information Systems: Current Status, Future Directions.” MIS Quarterly 38, no. 3, pp. 809–830.

Cerasoli, C.P., J.M. Nicklin, and M.T. Ford. 2014. “Intrinsic Motivation and Extrinsic Incentives Jointly Predict Performance: A 40-Year Meta-Analysis.” Psychological Bulletin 140, no. 4, pp. 980–1008.

Daugherty, P.R., and H.J. Wilson. 2018. Human + Machine: Reimagining Work in the Age of AI. Boston, MA: Harvard Business Press.

Dulebohn, J.H., and J.E. Hoch. 2017. “Virtual Teams in Organizations.” Human Resource Management Review 27, no. 4, pp. 569–574.

Fildes, B., M. Keall, N. Bos, A. Lie, Y. Page, C. Pastor, L. Pennisi, M. Rizzi, P. Thomas, and C. Tingvall. 2015. “Effectiveness of Low Speed Autonomous Emergency Braking in Real-World Rear-End Crashes.” Accident Analysis & Prevention 81, pp. 24–29.

Gollwitzer, P.M. 1999. “Implementation Intentions: Strong Effects of Simple Plans.” American Psychologist 54, no. 7, pp. 493–503.

Goodman, P. S., and T. L. Griffith. 1991. “A Process Approach to the Implementation of New Technology.” Journal of Engineering and Technology Management 8, nos. (3–4), pp. 261–285.

Google. 2018. “Smart Compose Comes to Pixel 3 and Four New Languages.” Retrieved from https://blog.google/products/gmail/smart-compose-comes-pixel-3-and-four-new-languages/

Grant, A.M., and S.K. Parker. 2009. “Redesigning Work Design Theories: The Rise of Relational and Proactive Perspectives.” Academy of Management Annals 3, no. 1, pp. 317–375.

Greller, M.M., and D.M. Herold. 1975. “Sources of Feedback: A Preliminary Investigation.” Organizational Behavior and Human Performance 13, no. 2, pp. 244–256.

Griffith, T. L. 1999. “Technology Features as Triggers for Sensemaking.” Academy of Management Review 24, no. 3, pp. 472–488.

Griffith, T.L. 2012. The Plugged-in Manager: Get in Tune with Your People, Technology, and Organization to Thrive. San Francisco: Jossey-Bass.

Griffith, T.L., and J.E. Sawyer. 2009. “Multilevel Knowledge and Team Performance in a Fortune 100 Technology Company.” Journal of Organizational Behavior 31, no. 7, pp. 1003–1031.

Hall, D.T., and E.E. Lawler. 1968. “Unused Potential in Research and Development Organizations.” Research Management 12, no. 5, pp. 339–354.

Hornung, S., D.M. Rousseau, J. Glaser, P. Angerer, and M. Weigl. 2010. “Beyond Top‐Down and Bottom‐up Work Redesign: Customizing Job Content through Idiosyncratic Deals.” Journal of Organizational Behavior 31, nos. 2–3, pp. 187–215.

Jesuthasan, R., and J.W. Boudreau. 2018. Reinventing Jobs: A 4-Step Approach for Applying Automation to Work. Cambridge, MA: Harvard Business School Press.

Kahneman, D. 2011. Thinking, Fast and Slow: Farrar, Straus and Giroux
New York.

Kozlowski, S.W.J., J.A. Grand, S.K. Baard, and M. Pearce. 2015. “Teams, Teamwork, and Team Effectiveness: Implications for Human Systems Integration.” In APA Handbook of Human Systems Integration.

Leventhal, H., R. Singer, and S. Jones. 1965. “Effects of Fear and Specificity of Recommendation Upon Attitudes and Behavior.” Journal of Personality and Social Psychology 2, no. 1, pp. 20–29.

Louis, M.R., and R.I. Sutton. 1991. “Switching Cognitive Gears: From Habits of Mind to Active Thinking.” Human Relations 44, no. 1, pp. 55–76.

National Science Foundation. 2018. “NSF’s 10 Big Ideas.” https://nsf.gov/news/special_reports/big_ideas/human_tech.jsp (accessed May 11, 2018)

Parker, S.K., F.P. Morgeson, and G. Johns. 2017. “One Hundred Years of Work Design Research: Looking Back and Looking Forward.” Journal of Applied Psychology 102, no. 3, pp. 403–420.

Plumptre, E.H. 1978. The Tragedies of Sophocles. London: Daldy, Isbister & Co.

Salas, E., S.W.J. Kozlowski, and G. Chen. 2017. “A Century of Progress in Industrial and Organizational Psychology: Discoveries and the Next Century.” Journal of Applied Psychology 102, no. 3, pp. 589–598.

Samuelson, W., and R. Zeckhauser. 1988. “Status Quo Bias in Decision Making.” Journal of Risk and Uncertainty 1, no. 1, pp. 7–59.

Schwab, K. 2017. The Fourth Industrial Revolution. New York: Crown Business.

Thaler, R., and C. Sunstein. 2008. Nudge: Improving Decisions About Health, Wealth, and Happiness. New Haven, CT: Yale University Press.

Toncar, M.F., and J.M Munch. 2003. “The Influence of Simple and Complex Tropes on Believability, Importance and Memory.” Journal of Marketing Theory and Practice 11, no. 4, pp. 39–53.

Trist, E.L., and K.W. Bamforth. 1951. “Some Social and Psychological Consequences of the Long-Wall Method of Coal-Getting.” Human Relations 4, no. 1, pp. 3–38.

Verghese, A., N.H. Shah, and R.A. Harrington. 2018. “What This Computer Needs Is a Physician: Humanism and Artificial Intelligence.” Journal of the American Medical Society 319, no. 1, pp. 19–20.

Wood, W. 2017. “Habit in Personality and Social Psychology.” Personality and Social Psychology Review 21, no. 4, pp. 389–403.