Finding and fighting the forces of churn in your organization

Trends come and go in the technology business. The thing today is often a forgotten relic tomorrow. I still smirk whenever I think about the arrogance of organizations like Pets.com or WeWork. The former was an example of bad business management and the latter fraud, but they were both examples of greed and hyperbole becoming a deadly mixture. That is why I take each big technology announcement with a significant grain of salt. History does not repeat, but it often rhymes as technologies fall in and out of favor. On the blog this week, I want to discuss a trend that should temper our enthusiasm about artificial intelligence in the office.

In the James Bond novel "Goldfinger," Ian Fleming said, "Once is happenstance, twice is a coincidence. Three times is enemy action." Before becoming a novelist, Fleming worked for the Royal Navy as an intelligence officer, so he knew something about enemy action. Over the last seven days, I noticed some articles that pointed out artificial intelligence's power and the challenges it creates for organizations.

AI and Code Churn –

The first comes from Microsoft and Visual Studio Magazine, the one-stop shop for news about developing with Microsoft tools. They noticed an interesting thing when software development teams use AI tools like CoPilot. Microsoft claims that providing developers with AI tools leads to declining code quality. While the number of commits and pull requests to Git repositories rises, these actions don't translate to improvements in code quality. Much of the software code inserted into project repositories is copied and pasted from CoPilot, jammed into existing code, and made to work. It is up to more experienced and professional developers to then refactor this copypasta code into a project. It creates what the authors call 'churn.'

On average, git repositories see a 55% increase in check-ins and pull requests, but there is no similar increase in releasable value in production. I liken it to a baseball team that gets plenty of hits but cannot get players to score and lose games. Each player has a respectable batting average, but the team is lousy and losing baseball games. I am a fan of the Chicago White Sox, so I have been familiar with this pattern since 2005.

Developers disregard established principles like code redundancy avoidance and dependency injection when working with AI co-pilots. It happens because the AI assistant cannot comprehend and apply these principles within the broader program context. Whether tasked with generating a bubble sort function or constructing a structure from a JSON string, an AI tool strictly adheres to its prescribed instructions. It shows me that Artificial Intelligence does some incredible things but needs more wisdom, judgment, and experience that humans provide naturally. CoPilot can spit out code, but humans still have to figure out if that code will solve a business problem.

Agile and Churn –

Artificial Intelligence is creating code churn, and the frantic nature of the global economy creates situations that create inefficiencies in every organization. According to the Harvard Business Review, the demands of the business world and the misuse of agile business practices are creating paralysis in organizations. After a month of effort, management's work is disregarded as executives flip-flop on their direction. Forced, last-minute direction shifts create organizational gridlock by robbing employees of the control and stability needed to innovate. When their efforts are constantly redirected with no tangible progress, employees become discouraged and unable to reach their full potential.

I remember working at an office where a project manager cynically muttered during a meeting, "Everything is twice as nice because we have to do it twice." The woman reflected on the needless rework and frustration that comes with projects that do not have clear ends, so business people focus on means instead. For the people who do the work, it feels like being on a treadmill with no forward progress, and it kills morale and innovation.

So, if leadership needs to get the most out of agility, they need to understand the end goal they wish to reach instead of the effort to achieve it. It returns us to the notion that outcomes from work are more important than the outputs of work. The Agile Manifesto says, "Working Software over comprehensive documentation." We should respect that value.

Organizations unsure how to use AI creating churn–

Finally, Venture Beat pointed out that organizations need help to figure out how to leverage artificial intelligence. It makes sense because medium and enterprise-level organizations have risks that individual computing people do not. The biggest concern is security. Large Language Models and artificial intelligence might use that information to train more extensive systems. Thus, other organizations might indirectly use your data in their AI systems. It means that any proprietary information you might have will make the AI systems of your competition more powerful.

I was excited about using AI and large language models to spot fraud in a bank payment system. I had a senior consultant veto the idea because they were concerned about privacy laws and data security violations. We could use artificial intelligence to spot trends in payments and possible fraud, but we would have to maintain that data in a place that does not contain personal data. If the system did spot fraud, we would have to reverse engineer the transactions to a different data set with personal information. Regulators call this PCI compliance a necessary protection to prevent identity theft in the banking and insurance industry. It could also triple the project's cost before we wrote a single line of code. Business people hate this kind of uncertainty and expense.

Artificial intelligence requires specialized skills and has a steep learning curve. It means the data scientists, developers, and engineers needed to make this a reality are expensive. The human talent to create AI is costly, and the requirements for computing power are more expensive. The energy bill alone is frightening, and my efforts to develop an instance of AI in Microsoft Azure gave me sticker shock. The Venture Beat article discusses these realities and points out that larger organizations that can afford it have more success incorporating Artificial Intelligence into their business processes. Companies with talent and money are spearheading the early stages of the AI transformation. Other companies need human or material capital to compete.

Fighting the Forces of CHURN –

Connecting the data points of these three articles says that we are in a temperamental phase of technology innovation. Now more than ever, businesses need professional developers, executives, and agile professionals to help navigate this landscape. Wisdom and experience will become more valuable because, as we see from Microsoft AI tools, it is possible to generate more software code but not better software. Business leaders who fail to grasp the overall goal while impulsively shifting priorities abuse Agile, rendering it ineffective in achieving their objectives. Finally, Artificial Intelligence is gobbling up skilled labor, and it is expensive and challenging to leverage this technology in any business.

It means that leaders must learn how to understand better their organization's long-range goals instead of the minutia of how the business works. Businesses will pay more for developers with a sense of craft and professionalism; otherwise, they will pay more money and time reworking existing code in their systems. Finally, businesses must decide whether to invest in Artificial Intelligence because the costs and commitments are significant. It is the enemy action that Ian Flemming talks about.

These three trends can be a risk for your business, and together, I would put them under the general umbrella of churn. It is the opposite of a flow state where everything is disjointed and filled with paralysis. Organizations in a state of churn are like ships stuck on a stormy ocean. You might survive to sail another day or sink to the bottom of the sea. Having intelligent and experienced people with wisdom and insight will give you better chances to sail another day. Artificial intelligence is not guaranteed to do that.

Until next time.