If you’re running or managing an AI project, you may have found that you need to level up your skills and understanding of terminology. As a foundation, general project management skills are a great foundation for managing schedules, resources, and the people needed to meet organizational goals. These skills come in handy managing web projects, application development projects, even typical analytics projects, or maybe even nontechnical projects, running a marketing campaign or even building big construction projects.
General project management approaches are geared towards dealing with non-technology specific project management concerns such as: project integration management, project scope, time management, cost management, resource management, communication management, and procurement.
However, these general skills lack the specifics to successfully handle the rapid pace of AI change. AI and data projects have their own unique challenges that need to be understood and addressed where general project management falls short.
Beyond General Project Management – What is the role of project manager in an AI project?
AI projects are primarily data projects, and they have their unique challenges. And the rate of failure for AI projects is very high. Around 70-80% of AI projects that get started either never complete or complete to a point where they are not meeting the objectives and they fail and they have to be canceled. It’s a lot of time and resources wasted, but also some potential big risks because failed AI projects could mean lawsuits and could mean getting into trouble with authorities.
These failed projects can mean eroded trust with your users and customers. AI projects require constant attention. Unlike traditional projects, AI projects are never a “set it and forget it” endeavor. Generative AI has shown us why you can’t just “set it and forget it” with AI. What works today most likely will not work tomorrow. Even the very same model, using the same data or the same set of generative AI prompts, might give you completely different results just days apart. You can quickly go from acceptable to unacceptable results fast which can have a real negative impact depending on the situation.
It’s important to note that your data can constantly change. Your models are going to constantly change. And so, if that’s the nature of these projects, then that’s going to require AI project managers to really be adept in data specific methods that will adjust to these evolving requirements and also that will maintain flexibility in project approaches. And this is actually the case with all AI projects. Data drift and model drift will happen. And since AI projects are really data projects you need to constantly be paying attention to and managing your data.
AI projects demand ongoing attention and adaptation. Mastery in data and resilience to change are key for AI project managers to thrive in this dynamic environment. What works today will most likely not work tomorrow, and vice-versa.
The need for Trustworthy AI
Responsible, ethical, and trustworthy AI is no longer a nice to have; it’s quickly becoming a requirement. If you are going to be spending all this money, time, and resources on your AI projects, you want people to actually use the AI solutions. In the “early days” of AI even just a few years ago people weren’t thinking about AI projects this way. Ethical and responsible AI systems were a “nice to have”, not a “must have”. But now, people and companies are pushing boundaries of what is possible across all seven patterns of AI.
This necessity of trustworthiness is unlike most other projects. If you’re building a website for an organization, trustworthiness needs do not usually come into play. Yes, the site needs to be safe and secure, but ethical and responsible use of the website by an average user does not enter the conversation. Or if you’re building a mobile app, ethical and responsible use doesn’t come into play. However for AI applications, these are now core parts of the conversation. Trustworthiness really is a core part of AI projects, and that makes AI projects unique from other projects.
As a project manager, you now are responsible for the trustworthiness of your AI application. And of course, in the new legal and regulatory environment, there may be some penalties.
Understanding Best Practices Project Management Approaches for AI
So we need to add more to our general project management approaches. Think of project management skills as the base of our soup. Let’s add some more ingredients that are more specific for AI and data projects.
As shared in Cognilytica’s AI Today podcast on this topic, having a foundational understanding of data and the data cycle and the various approaches to managing data is important. Why? Because data is at the heart of AI. Machine learning is powered by trying to do inference on data. That means you need to train the system with data. So AI projects are all about the data, not about the code.
As a project manager, it’s therefore important to have knowledge and skills that are focused on the methods and approaches for dealing with data, especially in these environments of uncertainty and dealing with constantly changing data and model lifecycles and because AI systems are so sensitive to all aspects of data.
AI project managers are focused on dealing with the complexities of data and managing the highly iterative and constantly changing requirements to allow you to deliver what you want to deliver, to be successful while being flexible. You need knowledge about data sources, preparation methods, and data quality management.
Additionally you need to understand AI approaches. At a high level, you need to know aspects of model development, including the sorts of machine learning approaches and the needs for model evaluation and testing, and also approaches to model operationalization, and specific needs around data engineering.
AI projects move fast. Your focus as a project manager is to move towards short iterations. What you as a data and AI specific project manager is an approach to delivering successful AI projects in the face of constant, rapid change iteration.
The Cognitive Project Management for AI (CPMAI) Methodology is increasingly being adopted as a vendor-neutral, best practices approach for AI project management. In this approach, you start with your business understanding identifying what problem you’re trying to solve. Then move to data understanding that’s focused on the specific pattern(s) of AI that you’ve identified in step one, followed by gathering the necessary data you need and then prepping that data. Only then are you able to actually build the model, and follow on by evaluating the model with real world data (not fantasy proof of concept data), and “operationalizing” the model in such a way that it is constantly monitored and measured.
Approaches like CPMAI help project managers logically and successfully manage AI projects making sure project scope is defined and manageable. “Think big, start small, and iterate often” is the best approach for AI project success and allows project managers to have successful AI project outcomes.
