A Hitchhikers guide to AI in Project Management

AI projects are coming - here's a quick guide to help Project Managers take the bull (Artificial intelligence) by the horns.

Feeling a bit hazy around what exactly do those two letters mean for the future of project management? The AI wave has been building for quite some time now and with senior leaders scrambling to get necessary resources in place to leverage this technology, where does the project manager fit into the equation?

With experts labeling AI as a “capable assistant to project managers” and project professionals themselves, expecting the proportion of projects they manage using AI to jump from 23% to 37%, we can expect an imminent impact in the project management sphere. Whether you’re ready to embrace it with open arms or just trying to get by, it is essential to understand what this technology means for you and how it could contribute to increased project success.

With Gartner predicting full automation of 69% of a manager's workload by 2024, what seemed like a technology of the future is edging that bit closer to our doorstep. To help project managers who are a little lost around the concept of AI, we have prepared a list of the main concepts every project manager must get their head around to adapt to this changing landscape.

Artificial Intelligence VS Machine Learning VS Deep Learning

The first step in your journey involves gaining a solid understanding of what exactly Artificial Intelligence is and why it shouldn’t be confused with other buzzwords under this umbrella, like Machine Learning (ML) and Deep Learning (DL).

According to Andrew Moore, “Artificial intelligence is the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence”. At the moment, AI systems are designed to perform one cognition task really well rather than actually think for themselves. These cognitive tasks include learning, problem-solving, and pattern recognition.

Machine learning is a type of artificial intelligence that does what it says on the tin. It processes large amounts of data using statistical techniques and then learns from that data without being explicitly programmed. Simply put, ML can do things that human brains are not designed to do. It has a massive calculation capacity, and a systematic approach to the data, meaning it surfaces links or anomalies that a normal human brain would not spot.

Already feeling a little lost? An example might give you another perspective. A simple application of machine learning in AI is your spam mail, which would otherwise require manual filtering to avoid unwanted promotions. Google estimates that AI-powered filtering prevents more than 99% of spam from entering your mailbox

The final term that has really created the recent surge in AI use is Deep Learning which is an evolution of ML. Deep learning structures algorithms in layers to create an “artificial neural network” that can learn and make intelligent decisions on its own. These Neural Networks are inspired by the functioning of the human brain and enable AI systems to learn from the data presented to them. Contrary to ML, deep learning can determine on its own if a prediction is accurate or not through its own neural network. DL requires a huge amount of training and data to really create added value but when it all comes together, deep learning comes the closest to mimicking a real human brain.

Deep learning is where we see AI most prominently used these days such as in Tesla's self-driving cars. This rise in DL only became possible due to the phenomenal increase in calculation capacity as before then computers were not able to process such large amounts of data. Therefore, we should not be putting ML versus DL, rather it's the evolution of characteristics that show how these two processes differ.

  • ML places emphasis on the algorithm whereas DL places emphasis on the data
  • For ML, if the algorithm is flawed then the result will be flawed whereas for DL if the data is flawed then the result will be flawed.
  • ML uses algorithms to learn from data whereas DL structures algorithms to create an “artificial neural network”
  • ML makes informed decisions based on what it has learned whereas DL makes intelligent decisions on its own.

When we bring AI back into the mix the following graphic represents how all of the systems combine together.

The Data Imperative in the functioning of AI

Now that you have a grasp of these three distinct concepts that people often use interchangeably, let's find out why data is king with artificial intelligence.

The saying “you’re only as good as your weakest link” comes to mind when talking about data, because in order for AI to make smart predictions it needs clean, high-quality data. Going forward, ensuring data is consistent will be a key role for the PMO. Tasks such as ensuring there are no duplicate data sets or questioning what exactly numbers inputted or outputted represent will also become a major task for technical PMs. These are time-consuming details but reap huge rewards if executed correctly.

Another key point to remember when embarking on your AI adventure is that AI is not a magic wand: “if your process is out of control, adding AI to it won’t fix it”. This is why initiating with a data-first approach is so important, project managers must have full control over the process. In simple terms, the goal of AI is to firstly remove the busy work (essential but not value-added) and secondly obtain more accurate projections for the future. But it will only be effective if the algorithm has enough data to build the right logic (i.e. AI needs a ridiculously large amount of data). Your dream system will be able to eliminate the busy work while giving real-time data on the progress of your project.

AI is a team sport

Being aware of this essential step is a great start in getting to grips with where AI is going to affect a project manager’s everyday life. Project managers (PMs) must also become familiar with key roles such as the data scientist, the data engineer, and the infrastructure engineer to stay ahead of the curve when AI systems really start to ramp up in this domain. PMs are very likely to run into the issues mentioned above and this is where you may cross paths with the following roles:

  • Data scientists (sometimes referred to as machine learning scientists) are involved in the construction of machine learning models. They are responsible for the initial inspection of the datasets and can often identify minor trends and patterns within the data. The data scientist will be your go-to person for solving real project challenges as they can generate reports that are used to draw inferences.
  • Data engineers (also referred to as ML engineers) build the software and some of the infrastructure required for machine learning products to operate. Data is made available to the data scientists from the data engineers. The final machine learning model is prepped for real-world problems, thanks to the overarching architecture created by the data engineers.
  • While the data engineers prepare some of the infrastructure, the foundation is prepared by the Infrastructure engineer. Project managers work alongside the infrastructure engineer to scale ML initiatives. The infrastructure engineer often coordinates with data center partners to ensure operational excellence, from the geographical location of hosted data to hardware.

AI entering the Project Management Sphere

To give you a little more clarity on the scale of AI today in project management and where we see it going, we’re going to delve into some practical examples of AI at work. The main areas where AI exists today is in:

  • Automated cost and schedule updates, re-forecast, and notifications/reminders. Using predictive analytics, AI systems can determine when the actual project completion date might be or the likelihood of a team delivering a project on time. Example: Pharmaceutical companies are researching life-saving medicines in a fraction of the time and cost it traditionally takes by using AI to speed up the regulatory approvals market for the latest molecules.
  • Analysis of patterns in data. Example: AI-enhanced data visualization helps identify bottlenecks and areas where process improvements or other changes can be applied to improve overall performance. In the automotive industry machine learning applications are being used to analyze warranty data to identify safety or quality problems in automobiles and other manufactured products.
  • Providing project team basic insights/recommendations on suggested actions. Example: AI can provide insights into potential project risks related to compliance and cybersecurity. So if the project manager was inputting data, AI technology would be able to confirm if any personal information inputted is in compliance with regulatory standards or inconsistent with the data normally filled in for a specific field.

The areas we may see AI in the future include:

  • Interpreting Project Data (surfacing specific information). By applying AI to project data you will be able to detect early on any issues as soon as you have the slightest sign that something is off track. AI will introduce accuracy and precision in surfacing potential constraints to project progress.
  • Analyzing project data with more accurate forecasting of budget, schedule, and risks. In this Ted Talk by Tom Gruber, he describes how AI will eventually help us overcome our cognitive limitations and help us get smarter about projects by learning from our past failures and successes and also those of colleagues. Imagine if you could summon every bit of data from past projects and make decisions based on tested and proven methods!
  • Optimizing resource allocation of teams. AI could be very useful in identifying employees with the skill sets necessary to perform a task based on experience and capabilities. Alongside this AI will be able to use past data to determine capacity levels of employees on a project based on past performance and the employees' current schedules.
  • Customized aid to employees using predictive analytics. Human behavior is often the most difficult resource to track on a project however AI-enabled project management systems have the potential to make this behavior more . Offering personalized coaching based on learning habits, identifying when someone is doing something out of compliance, and even accounting for conflicts with a remote employee working in a different time zone.

Understanding the functioning and flaws of AI

The main challenge posed for project managers will not be adapting to AI, but rather learning how to use the technology in the right context, and to challenge the results through good knowledge of the risks and limitations (brittleness) of the system. Since AI is wholly dependent on the quality of the data it is fed, understanding the potential bias coming from inaccurate datasets will be essential for project managers. If there is a bias in the data used to train AI/ML algorithms, then that bias will be present in the decisions that the algorithms make.

As a Project manager, you will need to treat AI-driven decisions like picking your route on a GPS, constantly questioning whether it has given you the fastest route and if it has taken into account the different factors like traffic and road surface. Blindly trusting an AI’s decisions may actually create project chaos rather than improve the efficiency of a project.

Another significant hurdle is that powerful ML models tend to be very difficult to explain which leads us to our next flaw, the Black Box issue. The AI algorithm produces insights based on a data set, but the end-user has no insight into the logic used to reach this decision. This “black box” issue can be problematic in highly regulated industries such as financial services, in which regulators insist on knowing why decisions are made in a certain way. Already we have seen cases of this issue in real life with Apple co-founder Steve Wozniak noticing that the Apple card gave him a credit limit ten times higher than his wife's, despite the fact that the couple shared all their assets.

Common sense is something we take for granted as humans but for AI systems it still remains a distant possibility. Modern AI is designed to tackle highly specific problems, in contrast to common sense, which is vague and can’t be defined by a set of rules.

So what’s next for you? One of the best ways to ensure that you are constantly evolving as a manager is to be pre-emptive with new technologies. In a world where 40% of workers spend a quarter of their workweek on manual tasks, developing a level of preparedness for more automation could be the key to convincing yourself that those two seemingly complex letters aren’t so bad after all.

Join the AI Conversation

Education is the most powerful weapon which you can use to change the world

— Nelson Mandela

As much as this is a guide for Hitchhikers, we want you to progress to let’s say… someone who picks up the hitchhikers. Therefore, to really embrace this technology, we recommend getting involved in the conversation. Below is a list of some blogs spearheading the AI revolution in project management:

And the following resources include videos, upcoming webinars, and training courses that give you the possibility to be the innovator rather than the skeptic in your organization.