How data integrity will make or break AI project management

When ChatGPT was introduced in late 2022, the wheels in every tech leader’s head were turning. It was, and still is, obvious that generative AI has unstoppable potential to change how we do almost everything. Yet still, in 2024, leaders struggle to determine exactly how AI can be integrated into their businesses to yield reliable results. 

We talked to 15 tech leaders to get their thoughts on where their team is at with AI, and here are 3 things leaders are realizing about AI: 

1. Leveraging AI is inevitable in running a competitive business.
2. AI requires data to run. 
3. Poor data quality is disastrous for AI. But, most organizations lack data integrity.

In this blog, we dig into the data integrity issue related to AI, why it matters in project management, and how Zenhub is helping solve this problem.  

Lack of data integrity is the reality in most organizations–and it’s limiting AI

For businesses, leveraging AI means avoiding the possible pitfalls of AI–hallucinations due to lack of data and the proliferation of inaccurate data.  “Solid data and processes are essential to getting the most out of AI, i.e., avoiding garbage in and garbage out.” Eric, a VP of Information Technology, tells us. 

The problem? In surveying tech leaders, we found that how clean a team’s data is is often unknown, and poor data hygiene is too often normalized. In fact, 70% of respondents mentioned data quality and process alignment as one of the biggest challenges with trusting and implementing AI. 

“The organization wants AI because of the buzzword, and they hope that commercial tools will do the trick. Lack of data governance, organization, and structure are still missing,” says Nuno, Head of IT Applications, IT Services & Operations. Nuno’s statement echoes many survey respondents–it seems most teams are hoping to find success leveraging commercial AI tools but are running into trouble with data quality in those tools.

Why data integrity matters in project management 

Like in most industries, AI is inevitable in project management: “AI is coming, and I think it will play a big part. If rolled out properly, this will provide major efficiencies within an organization. In theory, it should free up employee cycles to focus more on Agile working methods,” says Eric. These are big promises for a future free of administrative work that could be a major competitive edge, especially in the software industry where top talent is bogged down with mundane tasks. 

However, project management systems are databases. And what happens when these databases lack data integrity? One respondent, Chris, a Chief Information Officer, raised this concern and questioned if introducing AI would cause further “disorganization in an already loose Agile work environment.”

While there’s a clear advantage to AI project management, unclean data poses a risk–it could make mistakes such as referencing duplicate issues or using unstandardized labeling and formatting. These are all errors often made by humans that are subject to replication when AI trains on human-inputted data.

AI is well-positioned to reduce human error in data inputs 

Here’s the good news: right now, one of the jobs AI is most well-positioned to do is reduce human error in data inputs. At Zenhub, we’ve been thinking about this in two ways: 

1. Bringing uniformity to processes and formatting
2. Removing the number of human touches from task management to reduce error

Consider what generative AI is good at–picking up on patterns and replicating them–this is incredibly useful for bringing uniformity into data input. Once these patterns are recognized, they can be leveraged to reduce the number of human touches, reducing human error. 

Take Zenhub’s AI labels, for example. These are based on all your GitHub Issue data–while this data may be imperfect, it’s usually correct. By suggesting the label that is most likely to be correct, we can reduce the likelihood of a user choosing the wrong label (or no label at all), improving the uniformity of the data.

We’ve done a similar thing with AI acceptance criteria, using an existing acceptance criteria format, Behavior Driven Development (BDD). AI can take information from the Issue and reformat it to fit with the BDD format, reducing the likelihood that someone filling out and reviewing it will miss a critical component outlined in that format. 

The possibilities with AI are endless, but we need to start with data first

AI has the potential to perform much more impressive project tasks like training teams to improve their processes and driving continuous improvement across various functions. However, to unlock these advanced capabilities, we must first focus on ensuring the data integrity piece is correct.

The Achilles heel of AI systems is that they can falter without adequate data integrity, leading to inaccurate outputs and undermining trust in the technology. By focusing on AI use cases that can ensure more clean, well-organized data, businesses can set themselves up for even more success with AI in the future.

Tools like Zenhub’s AI-driven features demonstrate how leveraging AI to standardize and improve data inputs can significantly improve project management processes in the long run.

Interested in seeing how it works? Schedule a demo of Zenhub here.