TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Three Areas Where AI Can Make a Huge Difference Without Significant Job Risk

As we roll out AI, we are focusing too much on productivity and not enough on the things that truly need fixing.

• By Rob Enderle
• July 10, 2024

To me, it continues to look like we are focusing AI on the wrong things, such as automating things people like to do rather than things they hate doing. For instance, Microsoft’s Copilot can code for you but doesn’t automatically comment on the code, nor does it assure the quality of the result (yet). These are both things a coder typically doesn’t enjoy doing but are critical to the positive outcome of the result.

For Further Reading: The Question Everyone Should Ask Before Deploying AI Lessons Learned from Facebook’s Poor AI Implementation The Unfortunate Decision Process That Is Leading to AI Deployment Failures

Building ever-lower-quality offerings at ever-higher speeds feels like a going-out-of-business plan to me, so I’m going to suggest three areas where AI needs to be focused that would improve company performance without costing jobs and better set up the company for broader future AI deployments.

Quality Control (QC)

I’m a big believer in quality control. When I worked for IBM, my software division’s leadership concluded that we didn’t need quality control because development already knew about all the problems and QC slowed down the process. It shouldn’t be a surprise to find out all the executives making that decision were let go a year or so later because quality tanked.

Doing a QC job can be annoying because even though the job is critical to the outcome, your non-QC peers and management treat you like a potentially avoidable annoyance. You stand in the way of shipping on time and at volume, potentially delaying or even eliminating performance-based bonuses.

We are already discovering that to assure the quality of an AI-driven coding effort, a second AI is needed to assure the quality of the result because people just don’t like doing QC on code, particularly those who create it. When you are writing code at AI speeds, it isn’t clear a human QC specialist could even keep up with what is being turned out.

I’ve always believed that quality comes before speed, yet we are ignoring quality for speed, which has had some bad results at Boeing this year.

Efficiency

When I first started in business, there were a lot of efficiency experts (though they were in decline). The concept was someone who could look at a corporate process, working group, or organization and offer advice on how to optimize it for better results. In a way, analysts picked up some of this work as did specialized consultants, but the problem is that these people often don’t know enough about the client’s operations and team dynamics to provide viable advice and direction.

For instance, we know that within every company there is a core group of folks who make everything work. These are people who often operate outside what normally would be determined to be their duties to create working environments that are successful. We typically don’t make much effort to identify them and only conclude the criticality of their nature to the company once they leave.

This is particularly problematic when it comes to layoffs that are done haphazardly to get them done quickly. I was involved in one layoff where we ended up having to hire back a lot of people because we’d lost critical mass in manufacturing and could no longer build the product. We didn’t understand the critical nature of the employees we fired and had to hire them back at a considerable premium, undoing much of the savings the layoff was supposed to achieve. We didn’t even track relationships. I recall we didn’t realize there was a relationship between one of our largest customers and an employee which bit us hard when that employee was forced to leave.

AIs could monitor most employee-to-employee and employee-to-management interaction, scan their work product, identify problem areas that need to be addressed, and identify employees who are inefficiently placed (who are either over or underutilized), need help, or are working against the greater good, either on purpose or accidentally.

In short, properly applied AI could highlight and help address problems that are critically reducing a company’s ability to perform to its full potential and preventing it from becoming a great place to work.

Compensation

One of the huge problems companies of any size deal with is the unfairness of compensation. The problem arises because there is no truly objective mechanism that can assure equal treatment in any company. There is no argument that people shouldn’t be paid what they are worth, but we simply don’t know what any employee is worth.

Calculating an employee’s contribution and then using it to set compensation transparently should significantly reduce the number of employees who feel they are being treated unfairly by eliminating that unfairness or by showing them a path to improve their value and thus positively impact their pay.

In addition, it would highlight far better who the most and least valuable employees are, helping management identify under-performing employees (and take action to improve their value) and to provide a better guide for when and how the company needs to downsize.

Focus on the Fixes

As we roll out AI, we are focusing too much on productivity and not enough on the things we need to fix before we speed up our processes significantly, including quality control, employee efficiency, and compensation inequality. The last two are interrelated.

If we focus AI first on these three areas, moving to higher productivity won’t create (as it now seems to be on a path to do) even more problems for us by spiking unemployment, recalls, and other avoidable problems and expenses.

AI can help us create a better world, but only if we focus it on tasks related to that improvement. We don’t seem to be doing that right now. We should revisit our priorities before our existing performance-based priorities bite us in the backside.