Experiments in artificial intelligence

Project managers have heard time and again about how artificial intelligence (AI) is going to revolutionise the industry. But the reality is that project managers are just starting to consider the implications of AI, and the solutions that are really going to make a difference haven’t been built yet.

The latest prediction about the AI revolution comes from consultancy Gartner, which forecasts that 80 per cent of project management roles will be eliminated by 2030 as AI takes on traditional project management functions such as data collection, tracking and reporting. But the same report highlights that programme and portfolio management (PPM) software is behind the times, and AI-enabled PPM will only begin to surface in the market sometime this year.

“The market will focus first on providing incremental user experience benefits to individual project management professionals and then will help them to become better planners and managers. In fact, by 2023, technology providers focused on AI, virtual reality and digital platforms will disrupt the PPM market and cause a clear response by traditional providers,” Gartner’s analysts say.

Waiting for takeoff

Project managers are using AI today, as everyone is, in the apps on their smartphones and in smart office apps that schedule meetings, for example. Tools such as X.AI (scheduling), Aptage (forecasting) and Automated Insights (data to text) are already available. But these simple AI algorithms are not the shiny AI-enabled project management future that has been promised. Using AI to improve efficiencies in project management has been slow to get off the ground, for a number of reasons.

“If you were to Google ‘tech buzzwords of 2019’, you’d really struggle to find a list that doesn’t mention AI, and this hype has generated much enthusiasm in the project community. While the hype in this case is fully justified, because AI has revolutionary potential for project management, the reality is that most organisations are only just starting to develop robust data strategies and are currently in significant data debt,” says Alex Atudosie, manager of business intelligence and analytics at software consultancy Techmodal.

“AI is entirely dependent on the availability of large volumes of high-quality data, and this is currently a very scarce resource. In fact, a recent Harvard Business Review study shows that just three per cent of organisations meet the basic minimum for data quality, and I believe this is highly reflective of project data.”

The lack of high-quality data from which AI can learn and pull relevant insights is not just a problem within organisations, but between them as well. Martin Paver, founder and CEO of consultancy Projecting Success, is working on educating project managers about the art of the possible and concrete solutions to the problems of big data. In basic terms, what is learned during one project is siloed, and project managers at other organisations don’t have access to it.

“The job for the next couple of years is building out the data set and then thinking about use cases. People say: ‘There’s a load of data, let’s just start at it and see what emerges.’ But we shouldn’t be doing it that way. It’s got to be driven by use cases, thinking about the problem that we’re trying to solve,” says Paver. “If you look at a lot of projects, they’re the same, like building out Office365 for example. That’s being done over and over again. Or look at Crossrail. They’re not leveraging Crossrail data on HS2 or using it for Network Rail. It’s all siloed when it should be shared and integrated.”

Paver is trying to get project managers to contribute to a data trust, from which every organisation could extract insights to improve project efficiency across the board. To make sure they get the most from AI, project managers and project management organisations and institutions need to change their outlook completely, he says.

Grabbing the opportunity

That’s something that Greg Lawton, founder and CEO of AI solutions firm Nodes & Links, believes very strongly.

“I’d like to see a change of perspective on the UK’s position in the project world,” he says. “We’ve got this portfolio of around £600bn lined up, which is very big. But it isn’t big when you consider the size of the world and when you consider the fact that we’re now able to share data-driven understanding between projects and improve the core understanding of how they work – but we’re not doing that.

“The Infrastructure and Projects Authority is definitely set up to look at this, although it probably needs more teeth. How do we make the UK the world leader in complex project delivery? That opportunity is there. If we don’t do it, China will. If you’re producing 20,000 miles of roads a year, you’re going to be learning how to build roads very well.”

It’s also important to know the difference between how AI-enabled automation can change project management and how AI-enabled insights from massive databases can make a difference. Lawton gives the example of a client who was using people to count lengths of pipe. Turning that job over to an algorithm saved the firm 30,000 hours a month, a valuable efficiency, but AI has the potential to do much more.

“My advice to project managers is to understand how they create value within their business model and then to be open to talking about that and discussing this with the new generation of AI companies. Only then will the art of the possible be known. And only then will solutions come to fruition that aren’t just features that completely miss the ball,” says Lawton.

“We’re working with multiple projects across defence, aerospace, energy, transportation, infrastructure and construction, and what we’re doing is having that conversation. We’re asking: how do you create value for your client and how do you create monetary value for yourself?”

Realising the potential

Paul Taylor, executive technical director for programme management at professional services firm Stantec, explains that the business is experimenting with machine learning but is expending most of its efforts right now working on standardising its fragmented data sets to get data out of Excel and into a fit state to get AI processes working effectively.

“Once we’ve got that data sorted, we can buy in cognitive services from cloud companies. They can take the data sets and use AI tools to reanalyse them to give you predictions from the information you’ve provided. For example, you could look at resource analysis far more intensely or look at durations far more effectively. We’ve started testing that and have formed an in-house analytics team to build our own AI models. It’s proving very valuable.”

One project is looking at what triggers an alarm at a water treatment plant. “When it trips, is it something to be looked at, or is it only when three or four separate alarms trip that we have to get the operative out to look? The analytics team has been looking at that to improve the ‘find and fix’ work for water companies,” says Taylor.

He adds that robotic process automation (RPA) is not a priority for Stantec. “RPA is one of the toolsets that will replace tasks such as filling in timesheets. Instead of a human, a bot will fill it in. The cost of a bot is about £15,000 or £20,000, and then it will run 24/7.” He sees processes like this as simply augmenting the day-to-day work of project managers, but not replacing them.

Instead of RPA, Taylor is looking at introducing more fundamental AI processes through cognitive services. “Instead of replacing a specific function with a bot, let’s improve the base services with a better software package.”

These AI tools will help Stantec make decisions about how the projects and programmes are performing. “We are expecting to bring these types of thought processes in from April 2020 and offer analytic services like these to our water clients in 2022 or 2023,” Taylor says.

He adds that logistics in project management is one of the easiest things for AI to start looking at. “We are working with some other companies on this, where you tag the product with a code and, as it moves through the product life cycle, you get information about that product all the way through.”

Taylor also explains that analytics like this are already being used on construction sites to determine when products should be delivered and whether it’s more cost-efficient to stand teams down, rather than paying them to wait.

Be curious

Any project manager who wants to start thinking differently about what AI can do for them needs to start by understanding AI a little better, advises Liam O’Dell, a project manager who recently completed a master’s thesis on AI and project management.

“Project managers need to start to learn about data analytics. You don’t need to know it all, but you need to have an understanding of each of the areas you are dealing with. What the future holds is a lot of data. It is becoming more valuable than the individual or the organisation. If you’re not able to use and understand the data, you’ll be on the back foot.

“Having an understanding of coding, Python [a programming language] and so on is increasingly important. I subscribe to the theory that the project managers don’t need to do it all, but they need to be able to appreciate it all. They need to be able to know what they need.”

Projecting the Future: The fourth industrial revolution – data, automation and artificial intelligence

This is the first in a series of short papers by APM on the challenges shaping the profession’s future. It follows on from the launch of a discussion paper in June 2019 that set out plans for a “big conversation” in 2019–2020 about the future of the project profession.

Throughout ‘Projecting the Future’, APM wants to explore the questions that matter about the future of the project profession. We want to hear your views, ideas and case studies relating to questions such as these:

• Does the project profession have the knowledge and skills needed to deliver value to organisations as they transform and adopt new technologies?
• Can the project profession work effectively with technologists and business decision-makers to provide the necessary leadership?
• How is your work changing as a result of big data, robotics and artificial intelligence (AI)? Can you provide examples of organisations today that are adapting and implementing new technology to deliver their projects successfully?

We hope you will share your views on LinkedIn, Facebook and Twitter, or by emailing ptf@apm.org.uk. We are particularly keen to hear about case studies of projects that are making innovative use of data, autonomous systems or AI today.

apm.org.uk/projecting-the-future/fourth-industrial-revolution

The inside view from Silicon Valley

“There is a lot of hype around artificial intelligence [AI] – that it’s something magical, because our perception of AI is dramatically influenced by the entertainment industry,” says San Francisco-based Dr Michael Wu, chief AI strategist at PROS. “The way I define AI is that it’s merely a machine mimicking human behaviour. It has two key characteristics. One is that it has to be able to automate human decisions and the subsequent actions. The second is that it has to be able to learn and improve its performance over time. The learning aspect is all about getting the feedback data into the machine itself, so that it feeds the machine-learning process to update and improve the model.

“There are a lot of opportunities for AI to help project management. AI is most useful for automating mundane tasks that people have to do, like scheduling. Another area where it can really help with project management is learning to react to situations. Say you plan out a project and you allocate a certain amount of resources for this type of company or this type of project. This typically takes three weeks. An experienced project manager can normally make this kind of decision, but if you make that data available to the AI system, it will be able to make the same kind of recommendation.

“Tools like X.AI are already here, but whether the industry adopts them is up to the practitioner. Very often, people don’t adopt these technologies because of the fear of losing value. A lot of people are afraid that they may be replaced, and for certain kinds of tasks, they will be.

“People need to move onto something more strategic, more important and more meaningful, like solving problems that have not been solved before, where there is no data for the AI to learn from. You have to go and manage those projects and see what works and what doesn’t. Once you have done that a few times and have collected enough data, the AI can learn to take over.

“People have to learn to get comfortable with it. Ultimately, it should make our work lives more meaningful.”