Introduction
Over the last 15 weeks, I have learned a tremendous amount about human resources (HR), learning and development (L&D), and artificial intelligence (AI). This is this first HR-related course I have taken since 2015, so many of the terms and language were unfamiliar to me. That being said, this course has given me a new lens in which I can more clearly see the interconnectedness of people within organizations and the role AI plays now and in the future. In my current field, training baseball players, we are seeing a surge in the utilization of technology. The use of high-end technology to measure the flight of a baseball, the positions and speeds of the body, and the path of a swinging bat has opened the floodgates for those who know how to leverage that data to improve athletes. As baseball continues to progress, we are now seeing technology being used as an L&D tool. More specifically, the Cincinnati Reds began recording and transcribing every coach-athlete training session to evaluate and train their coaches. This aims to expedite the learning process for the organization, but there are implications of recording interactions that were previously private. After discussing specifically what the Cincinnati Reds are implementing, I will introduce an L&D framework that utilizes AI. Then, I will focus on the challenges inherent to L&D that make integration with AI both difficult and costly. Finally, I will close the paper by reflecting on the group action learning project we completed this semester.
Current State of AI in L&D
Early in the semester, we outlined the four different gradations of AI: automation, assisted intelligence, augmented intelligence, and autonomous intelligence. Automation is simply the execution of basic tasks and is not considered a novel way of doing things. Assisted intelligence assists humans in making decisions, but these systems are hard-wired and do not learn over time. These systems have their limitations, as they are making decisions based on information that is being inputted by humans—all who have their own biases. Automation and assisted intelligence systems are the main types of AI being implemented with L&D currently. Augmented intelligence assists in decision making while continuously learning. Autonomous intelligence can adapt rapidly and does not require human interaction to make decisions—this may never be fully accepted in HR, as that would take the ‘human’ component out of ‘human resources.’ (Rao & Verweij, 2017). Prior to this course, I somewhat assumed that most companies would be utilizing augmented intelligence and pushing toward autonomous. However, after getting a better understanding of the different degrees of AI and learning where technology is currently at, it is clear that those upper levels of AI are still far away.
Use of AI in Professional Baseball
Throughout baseball, the use of technology is rapidly increasing as the availability of relevant data has skyrocketed over the last decade. That data, and subsequent reports, have mainly been used to identify strengths and weaknesses of athletes. However, this year, the Cincinnati Reds, under new Pitching Coordinator Kyle Boddy, have created a new system aimed at helping coaches. In an interview with Baseball America, Boddy described the approach. “We now videotape every single session our pitching coaches have with our athletes, not just training and exercising, which we do, but also in one-on-one sessions where they’re breaking down the data” (Cooper, 2020). He cites football coaching legends Nick Saban and Bill Belichick who have administered similar practices for years. Following each session, a program Trint automatically transcribes the conversations and generates searchable documents. I reached out to Boddy for further detail on exactly how this is being implemented, the level of AI in use, and the implications of this level of surveillance on the culture of the organization.
Boddy informed me that Trint was not accurate in transcribing their meetings, despite using pro audio gear. This suppressed their ability to create a database to analyze speech patterns. The lack of quality transcribing software brings to light an important issue regarding utilization of AI. Without robust, high-quality data, integrating AI into a system will yield little value. In this case, until technology allows one-on-one athlete-coach interactions to be accurately recorded and transcribed, programs analyzing their speech cannot be implemented to further the development of the coach and to inform managers of who is a high-performer.
Despite transcription technology currently not being up to par, what are some of the implications when this can be executed? Upon gathering the data, systems can generate what words and phrases are common amongst coaches. Management then is able to study what verbiage generates the most meaningful adaptations in athletes. On top of that, the coaches that are more effective communicators can be identified and rewarded, or in opposite cases, punished. Independent of rewards or punishment, coaches will get increased feedback on their communication. To note, Boddy informed me that Reds coaches would ideally receive updates every 2 weeks. This can dramatically increase the rate at which they improve on the job. Phrases that are ineffective can be altered or discarded, while phrases that lead to positive adaptations can be more widely applied. Of course, it is important to note that each coach-player relationship is unique and that each athlete will interpret a coach differently, but it is likely that trends will emerge.
In the near future where Trint, or a similar product, works effectively, what impact will that have on the culture of the organization? More specifically, how will coaches and athletes feel about always having cameras watch and record their moves, and should they get a voice in implementing this system? There are valid arguments on both sides here. On one side, there is a privacy and trust component that will weigh heavily with coaches and players. On the other side, to drive organizational development, this system certainly will pay dividends. While the idealist in me believes that all coaches and players should have a voice regarding this recording system, the realist in me understands that management has to make these decisions (and many others) that do not have universal approval.
Regardless of what decision is reached and who has input, the effects it will have on the organizational culture strictly lie in the hands of upper-management. If seemingly non-consequential events are recorded and lead to over-the-top punishment, a deep level of distrust may emerge between management and those below them. Climbing out of that hole will set back progress for years. With that in mind, I emphasize the incredible weight that early decisions will have on the culture when greater surveillance is put in place.
Although the Cincinnati Reds were not able to realize the implementation of this system due to Trint limitations and COVID-19, the impact it can have on the organization is monumental. From an L&D standpoint, coaches will be able to learn from their interactions in a fast and meaningful way. Better coaches make better players, leading to a better on-the-field product and more revenue for the organization. However, there are privacy concerns with this system. Constant surveillance can lead to distrust, so it is crucial that management deals with the influx of information with great care and consideration. When decisions are influenced by this system, a level of transparency will greatly aide in strengthening the relationship between the management level and the coaches and players.
Introducing a Potential L&D Framework into Baseball Organizations
When the Chief Learning Officer (CLO) is designing learning structure for the organization, one important factor to consider is how strict the learning structure should be. In this case, strictness implies the level of autonomy employees have when choosing their learning goals. Employees that intrinsically want to learn new skills will be far more motivated than employees that do not possess that inherent drive. Typically, this should be identified in the talent acquisition process. Hiring employees that have a history of acquiring new skills and using them to improve their job performance will certainly be more valuable compared to hiring those who do not exhibit these qualities. Essentially, creating a culture where learning is emphasized will be far easier with a group of highly driven individuals.
After identifying and hiring these individuals, it is key that the CLO creates a plan that can be carried out. When developing this plan, the CLO needs to consider the following:
the extent to which both learning that has taken place within the organization and the tacit knowing and formal knowledge this learning has produced are potential sources for strategic advantage, and the learning implications of the strategic decisions that are made, including alignment, ongoing assessment of results, development of core competencies, and skills required for operational effectiveness (Yorks, 2005).
The framework I propose involves creating a database of all formal and informal learnings each employee possesses. From that database, net promoter scores can be deduced similar to what Netflix and Amazon implement. However, instead of suggesting a new show to watch or a product to purchase, suggestions for relevant learnings will appear. Typically, net promoter scores are used to gauge customer satisfaction, answering the question: “How likely is it that you would recommend this company/product/service to a friend or colleague?” (Reichheld & Markey, 2011). In this application, net promoter scores would be used to recommend continued education endeavors across the organization.
First, the resultant database can measure what sources and degrees of education employees at different levels have achieved. Over time, high potentials can strive to acquire relevant knowledge that exists at the upper levels of the organization. Over time, their learnings, contributions, and subsequent promotions (or lack thereof) can be more thoroughly analyzed. On the other hand, individuals that are not contributing as much to the organization can be identified. Two conceivable consequences of these findings are: the employee is not engaged in strategic learning due to lack of desire; or the employee is not directing his or her learning toward meaningful outcomes that drive organizational progress. With assistance from AI, the latter employee can be redirected and refocused to meet organizational goals. In this situation, the CLO may find that certain learning opportunities presented to the employees are not relevant to the organization or that there are shortcomings in the process that drive learning transfer. The main role that AI plays in this scenario is to shorten the feedback loop that exists between management and employees.
To tie this to baseball organizations, here are some potential questions: what degree(s) do workers possess? What courses do they have certifications in? What other fields have they worked in? How many years of experience do they have in each field/role? What independent projects have they pursued? How many languages do they speak?
Now, in practice, how would this model work? Let’s say an organization hires an ambitious intern looking to move up. Upon hiring, the database inputs the intern’s current skillset and experiences. Next, the net promoter score system can suggest both formal and informal learning endeavors the intern can take if he or she looks to advance within the organization. Suggestions can be tiered by level of importance and expectations can be set from managers. This system supports continued learning in a structured manner that drives organization progress and pushes employees to strive for constant improvement.
Challenges of Integrating AI into an L&D Framework
While there is potential for AI to exist in an L&D framework, it is important to note the limitations, costs, and challenges associated.
At this point in time, AI is not easily capable of creating learning materials which companies can disperse to their employees. From an L&D perspective, formal learning courses will have to be generated by either current employees or outside sources. A noteworthy limitation of creating learning materials for employees is that different workers will possess varying knowledge bases at the time of training. This makes tailoring content to the individual very challenging. If content is too easy, workers may feel that the training is a waste of their time and will disengage. If content is too hard, workers may be overwhelmed by the material, leading to feelings of stress and low self-efficacy.
Aside from the content of learning programs, the costs also must be factored in. If the organization is large enough to have a team dedicated to content creation directed at improving its workers, then learning opportunities can be more tailored to organizational needs. This may be costly, but could pay off long-term. On the other hand, money and time may be better spent outsourcing many continued learning endeavors.
As AI is implemented in an L&D framework, how will its effectiveness, or transfer, be measured? The 70:20:10 framework has been used by practitioners to develop L&D programs. In this framework, 70% of learning comes from challenging work experiences, 20% comes from social learning, and 10% comes from formal training. However, when measuring transfer, how does this framework factor in? Johnson et al. (2018) found that this model is not delivering on learning transfer for four main reasons:
(a) an overconfident assumption that unstructured experiential learning will automatically result in capability development; (b) a narrow interpretation of social learning and a failure to recognize the role social learning has in integrating experiential, social and formal learning; (c) the expectation that managerial behavior would automatically change following formal training and development activities without the need to actively support the process; and (d) a lack of recognition of the requirement of a planned and integrated relationship between the elements of the 70:20:10 framework.
By identifying some limiting factors on the current transferability of training, steps can be taken to fortify an L&D program that integrates AI. In the class lecture on learning transfer, we reviewed a model that places great emphasis on pre and post-program activities to support learning transfer. When implementing learning interventions, these pre and post activities drive home importance the organization places on the training. Thus, trainees will be further incentivized to utilize the skills they gained in their day-to-day work. The framework outlined in the previous section could be used to track the integration of the training to evaluate its value to the firm. If the return on investment does not match the firm’s goal, changes can be made to improve the learning process for future programs.
Review of Group Action Learning Project
Many of the points I discussed above were simply expansions of thoughts and discussions that I had with my group. During these discussions we openly shared our opinions on the lessons taught in class, and I more deeply learned from their thought-provoking perspectives. As we progressed through the semester, we found that unless an organization is very well-funded, the current utilization of AI on the L&D front is limited. Even then, the usage of AI is exclusively in the automation and assisted intelligence gradations. Many of these companies are just beginning to implement AI into L&D, so I presume progress will be steady as time goes on. One of the biggest realizations we discussed is that even though AI is a highly publicized topic, the notion that self-learning programs are going to soon replace millions of human workers is largely baseless. Humans are still needed to tell the programs what to do and how to think; this will be the norm for some time. As with all technological innovations, some jobs will be lost, but others will be created, and society will be better for it.
Conclusion
Although AI will play a critical role in how businesses run in the future, currently, its capabilities are limited. Firms using AI in their L&D departments are mostly automating tasks with some pushing toward assisted intelligence. In baseball, the Cincinnati Reds enacted a plan to use video recordings of players and coaches to improve communication, but limits in transcription technology squandered their early attempts. One potential way to drive a culture of learning within baseball organizations is to create a system that identifies the educational experiences that are consistent among high achievers. Entry-level employees can be given a wide-ranging list of different qualities that are present at senior ranks along with a guide on how to gain those highly transferable skills. That being said, acquiring those skills may be costly, and some firms may be in better positions than others to fund L&D. With that in mind, although AI is currently limited in the L&D field, over the years to come, its implementation will meaningfully speed up the efficiency at which employees and organizations improve.
References
Johnson, S. J., Blackman, D. A., & Buick, F. (2018). The 70:20:10 framework and the transfer of learning. Human Resource Development Quarterly, 29(4), 383-402. doi:10.1002/hrdq.21330
Yorks, L. (2005). Strategy Making As Learning. In Strategic human resource development (pp. 48-65). South-Western, Thompson.
Reichheld, F. F., & Rob, M. (2011). The Measure of Success. In The ultimate question 2.0: How net promoter companies thrive in a customer-driven world (pp. 45-60). Harvard Business Review Press.
Cooper, J. (2020, January 04). How The Cincinnati Reds Will Use Tech To Coach Their Coaches. Retrieved from https://www.baseballamerica.com/stories/how-the-cincinnati-reds-will-use-tech-to-coach-their-coaches/
Rao, A. S., & Verweij, G. (2017). Sizing the prize What’s the real value of AI for your business and how can you capitalise? (Rep.). Retrieved https://www.pwc.com/gx/en/issues/analytics/assets/pwc-ai-analysis-sizing-the-prize-report.pdf
Comments
Post a Comment