Research shows project failure rates continue to be alarmingly high. Many organisations try to mitigate this trend through the deployment of project controls such as project management offices, project management methodologies, registers and databased of lessons learned.
It is indeed true that project delivery is a complex problem. There are so many unknowns - it is impossible to completely define everything upfront. The changing business environment, with technological advances, global competition and regulatory pressures, all add extra complexity. Project delivery is further complicated as it involves people: it requires soft skills to deal with the conflicting needs of stakeholders, team members, vendors and customers.
To counter this problem, momentum is growing regarding the application of computer-based automation tools to mitigate project failure. As Yemmi Agbebi, Head-of Global PPMO & PD at Vodafone, said: ‘We know AI is coming. We are ready to ride the AI wave instead of being swamped by it.’
As you read on, we’ll discuss the current state of AI in project delivery and the likely market direction of AI project tools. Specifically, given every project is different, we’ll explore whether AI can really be used to improve project success. Indeed, we’ll go further and ask if the market is addressing the real project problem.
Project management - what’s the real problem?
The Project Management Institute defines a project as a temporary endeavour undertaken to create a unique product, service or result. Project management is the application of knowledge, skills, tools and techniques to project activities to meet project requirements.
There are many published articles on the causes of project failure, showing there are many factors that could impact time, cost, quality and scope constraints in a project. The project management problem could be summarised as: how do we remove the uncertainty from projects, or at the very least, reduce it?
Traditionally, a high degree of reliance is placed on human judgement and expertise to achieve project success, but this is prone to inconsistencies. With advances in AI, computers can now look to remove subjectivity and bring more rigour to the project management process.
Artificial intelligence
AI is seen as the new ‘silver bullet’ for all our technology woes; this includes increasing rhetoric pertaining to the project management problem. AI can be defined as the simulation of human intelligence processes by machines, especially computer systems.
These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions) and self-correction.
The adoption of AI is beginning to make its mark in industry. Gartner’s latest CIO survey of 3,160 CIOs from 98 countries, found that 21% of CIOs are already piloting AI initiatives or have short-term plans for them. Another 25% have medium or long term plans.
Evolution of AI in project management
The application of AI in project management is in its infancy, however. The diagram below shows the four phases of the evolution of AI in project management:
- Streamlining: improving existing processes through better integration and collaboration.
- Automation: the computer is actually “doing”; replacing human aspects of the project management process i.e. performing simple repetitive tasks.
- Insight & Foresight: the computer is assimilating project data and providing insights and recommendations to enable prediction of outcomes and better decision-making.
- Self-directed: the computer is autonomous; making project decisions and even remediating project issues that occur.
Over time, the level of human interaction reduces as AI in project management evolves. Many currently developed AI tools fall into the first quadrant (streamlining) with some now also appearing in automation and insight & foresight. However, self-directed tools (where the computer thinks, acts and does for itself) are on the horizon.
Streamlining: Integration and collaboration
There are 300+ project management tools on the market that perform a variety of tasks, such as scheduling, issue tracking, portfolio and resource management and team collaboration. Each tool has its strengths and weaknesses and organisations could potentially be using multiple tools.
The first phase in the evolution of AI in project management has seen chatbots being used to provide streamlined integration and workflow between applications. For example, chatbots have been developed to integrate project tracking and messaging tools. Integration enables the following efficiencies:
- Checking calendars for availability and scheduling meetings.
- Sending reminders for completion of timesheets, logs and other project data.
- Notifying project team members if sprints (set time periods in which a feature is developed and made ready for review) are likely to overrun.
- Generating project status reports.
Automation: Virtual assistants (bots) that perform simple tasks
The second phase is virtual assistants where bots are used to search through data created by projects to automate repetitive tasks. For example, bots can:
- Automate issue creation based on project team members’ instant messages. The chatbot searches through messages looking for discussions around potential issues. By automatic creation, the risk of issues being overlooked or swept under the carpet is reduced.
- Auto-check for data consistency and completeness. Data quality is a major challenge for project management. The lack of consistency between team members’ data entry limits project data value.
- Auto-check performance. User behaviour or project performance data is collected to determine where productivity can be improved.
Insight and foresight: Expert systems, machine learning and deep learning
The first expert systems were created in the 1970s and proliferated in the 1980s. They are rule-based systems, designed to emulate and replicate the decision-making ability of a human expert in response to an event or action. The system is composed of a knowledge base and inference engine; the inference engine applies logical rules to the knowledge base to deduce new information. The main disadvantage is the lack of creative response, which is where machine learning can help. Through training, algorithms improve their predictions, outcomes and responses over time as data becomes available.
Machine learning is a subset of AI which uses statistical techniques to give computers the ability to learn without being explicitly programmed. Statistical analysis of project data can be performed to gain insights to aid decision-making, which is magnified when machine learning is applied. For example, there is project planning software available that can learn from project history and create a regression model to provide future estimates of budget and task duration. Time, budgets and resources required to complete a task can be more accurately predicted by reviewing historical data.
Deep learning is a branch of machine learning that uses deep neural networks trained on large datasets and is particularly well suited for tasks like language modelling and text classification.
Self-directed
There are currently no use cases for this final phase. It is envisioned, as more of the decision-making is taken over by computers, that the role of the human project manager will change to focus more on strategic aspects.
Conclusion
AI tools for project management are still in the early phases of evolution. Currently, most tools are chatbot-based and focus on improving the efficiency of project management processes. While Integration and Automation improve the efficiency of project management processes, it will not be until the insight & foresight phase that project management AI will add the most value. However, this stage remains experimental, and we are some way off overcoming the project delivery problem.
