I am a small business consultant with 17 years practical experience in leadership within Fortune 100 companies in the fields of marketing, PR, profit and loss, change management, internal consulting, process enhancements and risk. I have a BA in Psychology and MBA. I will  complete a Master’s in Organizational Development and Leadership on April 25th 2019. My background has given me a perspective that is uniquely mine. I am fortunate enough to have a small business consulting firm to share my experiences with small businesses who can benefit most. Visit Strategy-Rocket.com to find out more information. 

 

I'm also a dog lover, Italian Greyhound aficionado and bird watcher.

  • Crystal Jones Taylor

Machine Learning: Bridging the Skill Gap for People and Organizations

Updated: Dec 7, 2018


Photo by Crystal Jones Taylor

Integrating machine learning and AI will require new skills, new organizational designs, effective QA and feedback loops. Additionally, the people, processes and technology must be integrated in a way that will create or sustain organizational wellbeing. Those organizations who can harness efficient organizational designs, QA, feedback loops and harmonious organizational deployment will have the sustained success we're all seeking.

As machine learning and artificial intelligence rock the business world, we must think beyond analytics, and instead, about a holistic approach to running our businesses, including the people behind the curtain. Specifically, how will we bridge the chasm between the traditional knowledge worker to a new paradigm where machines perform forecasting and predictive modeling in a more affordable and efficient manner? For example, organizational structures will shift as emerging skills reveal themselves and align in new ways. We should focus on future skills, as well as, workplace preparedness for managing the machine learning and AI impacts that will certainly send shockwaves through our organizational world. People will likely need to re-skill, and someone must bear the trade offs of time and money for doing so. But that's only one part of the equation. Let's examine people and organizations here.


People Implications:


We need to focus more on data integrity as the machines perform predictive modeling. How do our QA processes change, and what skills are required for validating accuracy in outputs? These are skills that will secure future employment. Task work is gone.

Companies that can best organize their resources most efficiently around machine learning and AI will hold the sustainedcompetitive advantage.

In times like these, employees often wonder if their skills will be become obsolete, or if they’ll be displaced altogether. We saw such worry during the advent of the computer. In the end, we saw job growth. Theoretically, this would assuage the fears of today’s knowledge workers. However, this history has done little to ease the worry among workers I know. None of us can be sure that robots won’t take over the world. In fact, I spoke with a massage therapist who said, “I don’t think a robot could do my job.” The operative word was think. He wasn’t 100% sure, like many people. I heard a fear inspiring radio commercial alerting CPAs that they should pursue a CMA (Certified Management Accountant) designation to demonstrate competency in decision making and strategy over quantitative CPA skills. In fact, some groups have been advocating the need for a universal basic income as a counter-measure to jobs displaced by machines. As you can see, fear abounds.

The types of jobs that seem most secure are subject matter experts with extremely deep domain knowledge, including fields like product management. Additionally, soft skills and the arts are resurfacing and becoming more valued. This is somewhat controversial: I believe that data mining, and to a certain degree, some amount of interpretation will become less necessary. I think only the most advanced business decision makers will be required to employ the data outputs. This means the data has to be extremely accurate, and thus the need for strong QA. Additionally, the computation must have a feedback loop to be able to accurately assess whether its predictions were correct. Without this, the machine becomes an echo chamber of false information. (See the NYT link below that shows the real-life implications of machine learning in the justice system as an example of this.)


Organizational Impacts


Identifying new talent is one of the tasks that machine learning has infiltrated most pervasively. We can learn from the Automated Tracking System (ATS resume filter) as one of the earlier machine learning/AI models. Being that the ATS can filter out age, race, gender, etc., it stands to reason that the workplace should become increasingly diverse. However, according to MIT researchers,

“Machine learning algorithms often work on a feedback loop. If they are not constantly retrained, they “lean in” to the assumed correctness of their initial determinations...”

I'll elaborate. Assuming that the ATS is successful at identifying the best candidates (and I have strong opinions on that for another time), we’ll need to continually recalibrate the ATS models to predict the success and retention of this newly identified machine learning skilled workforce. Are the skills we assumed were needed, like the CMA accountant, truly what we needed? Will CMAs be successful in their new roles? Further, do they stay? There must be a woman or man behind the curtain to make the final hiring decisions, right? With the ATS generating the ideal candidates (on paper) managers are currently making decisions around culture fit.


Making decisions around culture fit introduces its own complexities whose downstream effects impact the efficacy of the ATS. For example, if I’m the hiring manager and I happen to think I am a great culture fit who really knows how to get things done around here, I might hire a top contender with characteristics similar to mine. After all, it stands to reason that this candidate, too, should be a great culture fit. The hiring outcome feeds back into the ATS, suggesting that this candidate is the type to be successful. This hiring decision trains the ATS that this is the profile of the type of candidate we should seek out. The ATS becomes inherently biased. Hence, a diverse team should be building and tweaking the tool to (1) ensure it is not reinforcing its own biases in a loop and (2) eliminate the probability of unconscious bias in the programming. Once your dream candidates are selected, they must still be immersed in culture and be satisfied with their employer, overall. Additional resources must be designated, or redeployed, to create these fantastic employee experiences, as you cannot nurture relationships and retain your people resources via machines, or can you?


Integrating machine learning and AI will require new skills, new organizational designs, effective QA and feedback loops. Additionally, the people, processes and technology must be integrated in a way that will create or sustain organizational wellbeing. Those organizations who can harness efficient organizational designs, QA, feedback loops and harmonious organizational deployment will have the sustained success we're all seeking.

Here is an interesting article about how AI and machine learning are impacting the lives of US citizens: https://www.nytimes.com/2017/10/26/opinion/algorithm-compas-sentencing-bias.html


22 views
 
 

About Me

My blog spot for business and organizational perspectives

We are all told, “live your life to the fullest”; I am here to do just that. This site is a channel to engage in things that inspire me in this ever-changing world of business and the dynamic nature of organizations.

©2018 by Crystal Jones Taylor. Proudly created with Wix.com