graphic for The 2019 Index of Intelligent Technology in HR Tech

 

2018-06-01-hrexaminer-photo-img-cc0-via-unsplash-ant-rozetsky-272965-unsplash-article-can-hr-become-a-prediction-machine-rethinking-ai-and-talent-full-544x328px.jpg

“AI is best understood as a tool to lower the cost of prediction. And what are the consequences of cheaper prediction? First, more and more problems will be framed as prediction problems, because that makes the solution cheaper. Second, cheaper prediction increases the value of three key complementary assets: data, judgment, and action.” – Darko Lovrich

Can HR Become a Prediction Machine? Rethinking AI and Talent

 
This article is co-authored by Darko Lovrich and Dr. Tomas Chamorro-Premuzic.
 
Discussions about AI frequently center on technological issues, lost in buzzwords and arcana that may relate to “how” questions, but do much to obscure the “why” and “what” questions. In a thoughtful new book, Prediction Machines, Ajay Agrawal and colleagues at the University of Toronto provide a much needed antidote to the overhyped AI discussion that dominates most forums. The authors use classic principles from economics to demystify why AI is on the rise, what it means for leaders, businesses, and industries, and how organizations can derive more value from it. Their ideas have important implications for HR, which we discuss in this article.

2018-05-30-hrexaminer-photo-img-darko-lovrich-picture-sq-200px.jpg

Darko Lovrich is a Principal at Incandescent. Lovrich is focused on the people side of strategic transformation and innovation as well as advising start-ups in the people analytics, psychology, and neuroscience space.

The main thesis of Prediction Machines is deceptively simple – that AI is best understood as a tool to lower the cost of prediction. And what are the consequences of cheaper prediction? First, more and more problems will be framed as prediction problems, because that makes the solution cheaper. Second, cheaper prediction increases the value of three key complementary assets: data, judgment, and action. In other words, in a world where prediction becomes commoditized the main competitive advantage organizations will have is the ownership of data, the judgment to interpret the predictions derived from that data, and the ability to act and make decisions based on such judgments.

Applying their central thesis to the field of HR is straightforward. For example, recruiting is about predicting candidate’s performance in a certain job (not in absolute terms, but rather, compared to to other existing candidates). Development is the prediction of how much candidates will improve with coaching and training, or how much better they are likely to get. Compensation is about predicting attraction and retention at different levels of benefits packages. Workforce planning is the prediction of future talent needs in your business or industry, or a bet on how the job market will change. And so on.

While application is straightforward, we should note that predictive ability of HR is a gold standards few organizations presently reach. How can a company claim that they have a good model of leadership potential if they fail to predict how leaders will perform in the future? How can organizations be satisfied with their learning, training, and development programs if they cannot predict the ROI of such interventions? For the most part such predictions are rarely or never tested – unlike science, HR tends to subject its claims to rare falsification.

The second consequence of cheaper prediction highlighted by Agrawal and colleagues is even more instructive. When the cost of prediction decreases, to the point of being free, data, judgment and action become the scarce and most valuable resources. Prediction machines are built on and enabled by large quantities of high quality data. Without it, the old adage of garbage in garbage out rules. Secondly, while prediction machines can supply insights, these insights still require seasoned judgment to draw the right conclusion. And, finally, without taking decisive action on conclusions prediction has no value.

Imagine a superior predictive machine for talent management – call it FutureYou. It can identify employee potential with the highest accuracy, and estimate the likely performance gains you can expect from different interventions. Given the current state of affairs in employee selection and development, that would no doubt be a game-changer.

However, before you can implement FutureYou in your organization, you will need to ensure that a few key requirements are in place. First and foremost, you will need good data connecting employee characteristics with their performance, preferably over an extended period of time. Secondly, you will need to judge how much performance gains are really worth to you, and what your culture and organization is ready to do to bring them about, or even tolerate them. Finally, you will have to have access to a set of means for deploying them at scale if you really want to make a difference.

This example helps us tease out the three HR capabilities or assets that will become scarce as the cost of prediction continues to fall:

  1. Improving the quantity and quality of your data – Although we have spent much of the last 10 years hearing that data is the new oil and a source of competitive advantage in any field and industry, most organizations are still wondering how to collect it and use it. if you have big data on current or prospective employees and leaders, and you can link that data to relevant organizational outcomes, you are in a good place. If you don’t have sufficient data, consider investing in data/sensor systems that can help you harvest it within your organizations. Generally, you’ll need to be above a certain quantity and quality of data to make prediction work.
  2. Improve the quality of your judgment. For many organizations a set of vendors will likely provide prediction services – yet, they will still not be able to outsource the judgment required to act on these predictions. The ability to draw the right conclusions from the insights provided, and suggest the right interventions, will be highly contextual to each organization, and a key driver of your future business value. The good news is that you don’t need to turn your HR team into data scientists – it is sufficient if they are able to understand the main implications that different predictions have for their business, and they are able to make evidence-based decisions on that basis.
  3. Improve your ability to act. Prediction machines don’t quite focus on interventions – they are diagnostic tools.. You will need to add an intervention loop, consisting of: a) judgment about which data to focus on; b) judgment about which intervention to try; amd c) measurement of results. You will need to own this loop, and know its value not just as a tool to improve your organization, but also as an asset to enhance future predictions.  

Co-Author Bio: Darko Lovrich
 
Darko Lovrich is focused on the people side of strategic transformation and innovation as well as advising start-ups in the people analytics, psychology, and neuroscience space. In his work he combines strategy and psychology to clarify an organization’s vision and mission and to build a system that can execute that vision. Darko is particularly attracted to big missions and large-scale transformations. Prior to Incandescent, Darko was a Fellow of the World Economic Forum, where he worked with public and private sector leaders to understand our changing world, including deeper examinations of the impact of digital technologies on other industries, the future of health, the dynamics of resource scarcity, and the political realities of the Arab world. He started his career with Goldman Sachs as a Human Capital Analyst, subsequently joining Deloitte to focus on change management and post-merger integration. Both of these experiences provided insights into how people and organizations pursue (or avoid) strategic change.

Darko holds five university degrees (two bachelors and three masters) in the fields of psychology (Warwick), psychiatry (Oxford), business (Zagreb), foresight (Houston) and leadership (WEF in cooperation with Wharton, Columbia, INSEAD and LBS). He now feels he is credentialed enough and is actively avoiding further university applications.

Darko loves rowing, sailing, and restaurant-centered travel (which makes the other two hobbies necessary).
 

graphic for The 2019 Index of Intelligent Technology in HR


 
Read previous post:
HRIntelligencer v2.20

“Anyone with large organizational experience knows that the unpredictable is the norm. Contemporary AI ignores this reality, blithely assuming that...

Close