AI Implications on Project Management and has the Pandemic sped up the adoption?

Some articles have the genuine ability to inspire and spark the interest of the reader to want to learn more. ‘Brain Power’ in PMINet 2019 is such an article. In the article you could read about all the fantastic things that AI can do, now and in the future to support you as a Project Manager to run more successful projects. This by for example extracting valuable insights from lessons learned and get assistance in administrational work to allow you to focus more on motivating your team spend more time in dialogue with the stakeholders to ensure meeting the project objective. In the article, there were interviews with Geetha Gopal and Bruno Rafael Santos who shared real-life experience and valuable insights on how they have used AI in their work, which made the possibilities with AI in Project Management feel very tangible and within reach.

The same year, a Project Management Institute survey confirmed that artificial intelligence (AI) disruption is happening. In the article you can read that High-PMTQ* project management technology quotient organizations—and their project leaders—can see how AI technologies are already fundamentally changing the business landscape. In the survey you can read that eighty-five percent of global CEOs predict AI will significantly change the way they do business in the next five years, according to PwC. 1. New research from Pulse of the Profession® confirms AI disruption is happening: 81 percent of respondents report their organization is being impacted by AI technologies. And 37 percent of respondents say adopting these AI technologies is a high priority for their organization, sparking a shift in project management approaches. Over the next three years, project professionals expect the proportion of projects they manage using AI will jump from 23 to 37 percent.

So, what has happened 2,5 years later when a pandemic has dramatically changed the work life for the most of us? The natural place to work has changed from the office to our house and the meetings has mostly taken place online on Zoom or Google Meet and by that has sped up the technology advancement and forced us to accept new ways of working. Never has change management been easier and investments in technology more available. With all the digital advancements, are we ready to continue our technology journey and embrace and include artificial intelligence in our teams and our day-to-day way-of-working?

Since we believe that Artificial Intelligence and its potential impact on the Project Manager role and way of working can be a very important and interesting to our members and that there has been no follow-up survey published by PMI we reached out to Geetha and Bruno and asked them if we could do a follow-up on the interview from 2019 and if they could answer on the same questions from then. In addition, share their thoughts about how AI will impact the Agile way of working. We are so happy that they agreed to share their valuable experience and insights with us. 

In addition to this article, we feel privileged to announce that we will have Geetha and Bruno presenting at the Passion for Projects Congress 23-24^th May 2022 at Svenska Mässan in Gothenburg. Read more on Passion for Projects 2022 and book tickets already now. Early bird prices available until March 31.

 

Has the pandemic impacted the use and interest in using AI and Machine Learning in project management?

Geetha:Certainly yes. Several transformation reports have identified AI and Machine Learning as top investments during the pandemic.The remote and dispersed operating environment has increased the need to rethink processes and instead rely on data, machine intelligence and automation wherever necessary. Planning is essential for project management and most of the recent developments on the topic of AI for project management deals with intelligent planning, forecasting, tracking and allocating resources. The shift that we observe is that the pandemic has added another dimension to AI driven project management with virtual planning and remote-working dependencies. Given the high nature of dynamism and uncertainties we have faced in the pandemic, the key challenge that machine learning algorithms have faced is to incorporate pandemic data and predict meaningful outcomes for forecasting. On the other hand, the rise of intelligent communication tools driven by underlying AI technologies like Teams and Zoom have contributed favourably to virtual teams. Hence AI and machine learning have definitely had an impact in project management in several levels.

Bruno: First, we are still far away from true Artificial Intelligence in project management, and everywhere else. I say this because Artificial Intelligence is a mathematical theory still under development while Machine Learning is tangible already, but still primitive in comparison. Technologically, we still have many advancements to make in ML until AI arrives in a not so far future. Now focusing on ML, since 2019 many advancements in NLP (Natural Language Processing) were made like GTP3 and BERT, so potential applications in non-structured data (video, audio, text and image, a non-exhaustive list) are now available as commodities. Previous technologies like Deep Learning (neural networks) are heavily commoditized and easy to deploy in corporative servers, with cloud providers being very active all around (Google being the most prominent). So, tendencies in Machine Learning for Project Management are similar to overall IT tendencies: cloud, third-parties, open-sources, data sharing and privacy.
 
Second, the old challenge of data science, Data Literacy, is still on! The revolution of ML in project management is not a technological revolution, but a human, cultural, and behavioural one. Are we ready to use Augmented Intelligence (ML supported decision making)? Are project managers ready to use such technologies, understand their caveats, limitations and risks? Other social/cultural aspects have evolved a lot since 2019 due to the pandemic. Video conferencing, for instance, has been available for a while, now it is widely used and accepted, but it was so painful and distant some months ago. Privacy issues and cybersecurity are hot topics right now, and they must be taken into consideration when thinking about ML in project management. ML related project management (projects on ML development and deployment itself) are more common now than ever. Are the project managers ready to deliver under these conditions? Can they communicate with the consultants, developers, contractors, etc? This is a stakeholder management and team management challenge.
 
Third, some years ago if you need machine learning in your project you would hire a full-stack hacker to do it for you. Now we have a plethora of positions in this area: data architect, data engineer, data analyst, data scientist, Business Intelligence Analyst... The modern Machine Learning implementation is a team, not a person, this is a parallel to what happened in IT departments all over the world. The project manager must be ready to manage an ML team instead of an ML developer.
The Pandemic has impacted the interest in anything that can be automated, so using ML in project management activities is a natural strategy. The immediate impact of the pandemic was a communication crisis: day-to-day people-to-people contact disappeared in less than 24 hours. Setting up video conferences, scheduling meetings, considering multiple communication channels... Now simple communications must be planned, and simple tasks with documentation management have moved into the cloud. Properly managing communications and documentation are areas where project management still has to evolve and where ML could help by retrieving intelligence from the documentation. As I have said before, this is a Natural Language Processing (NLP) challenge in most parts.
 
There has been a trend for project management to become less a hard skill and more a soft skill. As the project manager becomes the project leader or disappears completely in some agile teams, the remaining activities under the project manager's control are human-related and not fully "machine-learning-able" yet.
 

What recent projects have you worked on where machine learning helped with decision making? is machine learning used more frequently in decision making now than 2 years ago?

Geetha: Firstly, the acceptance on machine learning and decisions based on them are improving. This also means that the need for generating clean and relevant data – to be able to use them in machine learning and guided decision making, is also gaining acceptance. In our day to day business requirements, we see our business units and project teams requesting to integrate AI/ML into solutions we offer. Due to this, we frequently find ourselves in situations where we have to engage in discussions revolving availability of relevant data.

Bruno:Still in 2019 we had the opportunity to explore the use of ML in risk management in a major project within the PMI Rio de Janeiro Chapter, long story short: in the previous iterations of this project (an annual event) we had had issues with some catering contractors due to overbooking, a common thing in events. We managed to create a simple model based on historical sales data to properly manage the size contracted instead of trusting our previous expertise. This reduced the costs with fees due to overbooking to zero and also reduced costs with other project materials such as credentials (collateral impact on ESG!). Another larger project in the same year had issues with contracts due to overbooking.
 
Right now we are preparing to deploy our Community of practice initiative, and, by doing so, create some new initiatives within the Knowledge Management Teams. We need to figure out a way to retrieve information from our database of presentations from previous events. We were able to make this into rectangular data employing taxonomy and classification techniques, but the raw data within the materials are still out of our reach. This is just the beginning, with the pandemic, the "Powerpoint Presentation" events are disappearing and giving place to videos, interviews, podcasts and other formats which have been difficult to summarize, annotate and tabulate. That is why our research towards Knowledge Management data retrieval is growing.
 

Is machine learning used more frequently in decision making now than 2 years ago?

Bruno: Undoubtedly, augmented intelligence is becoming closer to the market than before. However, the project management environment is not ML friendly at all. By definition "projects are unique endeavours created by people to create value", so they are very difficult to generalize. So creating an ML model for supporting decision making in project management is difficult by definition. ML behaves very well when dealing with processes, six sigma, lean, and anything that can be standardized. The classification works very well with a small number of classes, regression with predictable data, and so on. Projects in areas such as construction, industrial, supply chain, in other words, hard projects can greatly benefit from ML. Soft projects, where the output is less concrete, are very difficult to model (not impossible!) and models still need to be developed.

Obvious areas for the application of ML in project management are ones with rectangular data (numerical or spreadsheet-like data) such as costs, schedule, risks (the best candidate by the way!), quality and technical aspects within the project. Areas such as stakeholder management are still tricky; they can benefit from some Sentiment Analysis (NLP application) and scope management could explore Style Transfer (also NLP application). For instance, in sentiment analysis, we can evaluate the text from communications (e-email) and check if the stakeholder's sentiment towards the project is changing. In style transfer, a text is given to a model and it rewrites the text imitating some style (some author, or company). Most NLP technologies are similar to assistive technologies as these applications are the most demanded by the public. Such technologies can make project management more inclusive and accessible by consequence.
 

How can machine learning help project managers with resourcing needs? 

Geetha: HR is an area where AI/ML based decision making has been popular over the pandemic. Organizations have included AI-driven tools into their virtual hiring process and thereby heavily rely on machine learning outcomes for key decisions like considering suitable candidates, interview process, on-boarding, etc. Hiring to project teams have thus a direct impact through this shift. Though it is popular in tech and large organizations, this trend will continue to grow and spread across all industries and scales of organizations. 

Bruno: It can help in all stages, but may not be the proper tool for all cases. Let us begin with the budget! Analogous budgeting would greatly benefit from ML models that can quickly retrieve intelligence from previous projects, but do we have this information in a standardized, tabulated format ready to be used by the ML model? Do we have enough data to create a model? Does the model have reliable levels of generalization? This is a Knowledge Management challenge the some, if not most organizations, are still struggling to overcome.
 
There is still an issue. Budget data is mostly rectangular data and numerical data; that is the easy part. Non-budget related data, like WBS and Scope, are much less ML friendly. Here dealing with highly variable data is a challenge (graphs, text, image...). NLP tools would greatly benefit the project manager, also a good database of lessons learned to explore with the help of ML. These data structures may be so complex that ML may not be the immediate answer.
 
Sometimes what we need is Data Science itself, with basic Visualization, OLAP, Business Analysis, etc. Consider a large database of lessons learned. How does the project manager know which ones are related to the target project? Exploring the taxonomy (classification, description, dates, project titles, tags...) may reveal that there are several groups within the data structure: what are they? Projects? Industries? Clients? Do they mean something to the project manager? That is why a project manager must go beyond being a simple user of Data Science, that is Data Literacy. One must be able to use the basic tools in data science, enough to communicate with a technical team at least.
 
Also, some applications of ML are overusing. For instance, resource allocation is essentially an optimization issue. There are plenty of tools in Operations Research (an area in mathematics mostly related to decision making than ML and statistics) for this, which can provide a precise, fast and reliable answer without the limitations of ML.
 

How can project managers help ensure these tools filter the right data?

Geetha: Bias is a big risk in machine learning and AI. In most cases organizations procure tools and services and when they shift to such AI recommended decisions, it is not feasible to review or validate the biases in these algorithms. This is specifically an area where human intelligence is needed to complement the machine learned recommendations. In future we could have more reliable tools but in the current nascent stage, it is important for Project managers to engage in communications, look at references, seek clarifications and use the human factors in final decision making.
 
Bruno: Data literacy once again! The project managers should be at least versed in basic data jargon and have some domain knowledge in the areas they are working. That is how they ensure that the proper data is in place. Sometimes this is more challenging than it appears. An excellent example of issues arising from poorly filtered data is related to Structural Racism in the USA. Some ML models used in healthcare had severe racial bias due to poorly sorted data, the model's development and operation were excellent, but the data was biased, so the model was as well. This can nullify the benefits of a Diversity and Inclusion program within an organization, can create deleterious ESG projects and so on.
 
So all variables must be carefully considered: is the business object clear? Are the data collection strategies adequate? Are the data sources reliable? How are the legal aspects? Any privacy issues? Any ESG issues? Any social, cultural, ethnic and racial issues? Is this data human-related data? Does the model have any ethical implications? Which are the caveats of the related industry? Is it healthcare related? Is it education related? Each project is unique, that is the challenge.
 

How can machine learning tools help project managers most?

Geetha: ML tools can help in predictive analysis, intelligent and data-driven forecasting, identification of dependencies, potential risks, probability of occurrence of issues and risks, monitoring of the critical path, impact analysis, to name a few. They also help to improve discussions through data and facts, while increasing collaboration and communication.

Bruno:Difficult things, not easy things. There is a going meme/joke in the deep learning community about data scientists using Tensorflow (a programming library for deep learning development) to do Linear Regression (which can be done in Excel). This is not a joke! It is quite common to see overpowered tools being used in overly simple tasks. So let us focus on what ML can do for project management.
 
Budgeting? Maybe, it could be explored in some organizations where the projects have some similarities among them. If a model is not generalizable (each project is unique), it is useless.
 
Unconstrained Optimization? No, there are plenty of optimization techniques to be used, generalizing a model for Resource Allocation, for instance, would be too much. The existing tools like simulation and linear programming are good enough.
 
Risk Management. Yes, this is the natural candidate, the rectangular data, the predictive analytic, mature models and processes...
 
Retrieving intelligence from Knowledge Management databases? Yes, this is the hairier one, which non-structure data would be a challenge.
 
Stakeholder management? Yes, ML models could both help with stakeholder analysis and registry.
 
The raw impact of ML in Project Management and other areas is speed! ML can take complex data and make sensible decisions much faster than a group of human agents. This meansthat decisions, changes, predictions, prescriptions, and reroutes happen at light speed, and the project team must be aware and ready to answer to such demands as they happen. Also, organizations must be ready to answer properly, if an ML recommendation arrives in time but if the organization is not agile enough to exploit the opportunity then the intelligence was squandered. How much human intelligence is squandered daily? How much ML's intelligence shall be squandered per millisecond?

 
How can machine learning help analyse past projects to identify and assess risks?  Can machine learning help with the analysis when using Agile methodologies?

Geetha: Agile and AI have a unique connection. Multiple sprints and iterations of Agile tend to generate more data. Even within the same project, multiple sprints mean more data to train machine learned algorithms and make meaningful recommendations for the projects. Similar benefits can be expected while using data from multiple agile projects.

Bruno: Do we have a database of past projects? That would impress me! The data is complete, clean and updated? That would charm me! And also there is a tradition of lessons learned and a culture of knowledge sharing and collaboration? Wow!
 
We need lots of data to yield a simple, generalizable and reliable ML model. Let us assume we have the data. Are the projects comparable or are they oranges and apples? If they are too different, a model created from mixed data would be noisy and difficult to use. If they are comparable, which risk factors are time-related (if the factor is time-dependent, historical data would be trash)? The assumptions are still holding? If they changed, this can be quantifiable? Once we have a model, the model could easily support the decision making or suggest a list of possible critical variables to consider. All areas of risk management could be benefited:
 
Risk identification: from historical data, a list of possible risky variables could help the Project Manager to figure out which ones are the more hazardous.
 
Risk qualitative and quantitative analysis: what is the impact of such a risk, what is the probability? Has the given probability been reliable over time or has the prediction failed?
 
Risk response: which risks became concrete, which responses were implemented and deployed. Which produced the highest impact? Quantify the impact...
 
ML could do more for risk management than any other area. For instance, in some situations, the impact and the probability are estimations from marketing and external sources. With historical data, such estimates would be precise given the nature of the project or the organization. They could be tailored for each possible project condition and be more reliable.
 
Another area is Knowledge Management. From historical data, non-structured data, keywords and sentiment analysis could reveal risks related to stakeholders or reveal some risks that only existed in the reports, but not in the databases.
 
Can machine learning help with the analysis when using Agile methodologies?
 
Yes, it can. If we think of adaptive x predictive project management, the difference between agile methodologies and other methodologies is simply the implementation of some project management related tools. The Backlog is more friendly to ML scrutiny than most WBSs and Scope Statements. Historical data on Backlogs could easily retrieve interesting user stories that were left behind before they are considered by the team. Some agile time managementtechniques, mostly Kanban related, create data on the team productivity that could be used to predict variations in sprints and deployment.
 

How do you balance the benefits of machine learning versus your own project management instincts? 

Geetha: Intuition and instincts are skills that cannot be replaced by tools. On the hindsight they are subject to cognitive biases which could be driven by the individual's personality and preferences. On the other hand, data holds the key to bridge the gap between human instincts and machine intelligence. Like I always say, machine learning outcomes must be considered as guided decision making mechanisms with human intelligence to validate and finalize.

Bruno: Considering human-related stuff, the project manager's guts are still the cutting edge technology. So communications, stakeholders, negotiation, and scope are still the bastion of human intelligence in project management. Also, human intelligence is less effective with quantification but can explore complex relations among variables better than any ML model. The proper use of Exploratory Analysis can greatly enhance project management activities even in areas where ML has potential.
 
When quantitative analysis comes into play, things can change a little bit. ML models can be much more powerful in quantitative analysis and estimation than anyone. But if the models are not generalizable, they are not reliable, that is where the Project Management expertise comes into action. Also, considering the use and creation of models within a project, the PM expertise is pivotal in supporting the development of ML models that are reliable to PM instead of just powerful.
 

What do you say to those who worry machine learning might replace project managers? Have you noticed any change in how you distribute your time in a project as project manager when using machine learning.

Geetha: At this point of time, the savings of time from machine learned outcomes is negligent. What they offer is fact, data, insights and guidance to have meaningful conversations within project teams. Though we still have to have these conversations, we spend lesser time in manual tracking, evaluations, correlations, etc. Project managers must embrace data-driven project management to be able to spend more time in strategic activities, communication, stakeholder management and optimization. 

Bruno: No way! As I have commented in the first question. AI is decades away and ML is just a poor, but powerful, imitation. Even the most advanced models cannot surpass the human intellect. We must take ML as a tool to be used and Augmented Intelligence as its most practical and predictable future. But predicting the future of technology is quite easy, predicting the future of human behaviour is quite difficult. Some years ago we were moving towards a society without privacy, now everyone has privacy concerns. The phenomenon of the internet is human, not technological, the backbone of the internet is Telecom, a very predictable technology.
 
So, what is the future of machine learning? More commoditization! More ready to use models, more one-button applications, more powerful models, more powerful platforms, and more open-source projects. IoT (Internet of Things) is on the rise for retails. Healthcare and Fintech related revolutions are ongoing. How are we going to behave? Are we going to embrace cryptocurrency? Are we going to embrace wearable gadgets? Are we going to embrace Augmented Intelligence with ML or are we going to resist it?
 
Lately, I have spent more time talking to people and communicating than doing project management stuff like creating a schedule. I believe this has mostly to do with the agile mindset being more widespread than with any new technology or practice.
 
On the other hand, I have been automating lots of tasks lately (machine learning or not) and concentrating my efforts on human relations. What is still unclear to me is the cause-effect relation: Am I using more my soft skills because machine learning is helping me with dull work or am I using automation because I need to connect more with my peers? The second one sounds better and would be a more desirable mind-shift for all of us.

In addition to this article, we feel privileged to announce that we will have Geetha Gopal and Bruno Rafael Santos presenting at the Passion for Projects Congress 23-24^th May 2022 at Svenska Mässan in Gothenburg.

 

About Passion for Projects Congress

The congress is Scandinavia's largest meeting place for Project, Program and Portfolio Management. To make this year's event as succesful as previous years, we look forward to your participation. It is an opportunity to get inspiration, network and have fun. Several partners and exhibitors will be participating.

Read more on Passion for Projects 2022 and book tickets already now. Early bird prices available until March 31.

 

About the author

Marly Nilsson is PMP certified since 2011, volunteer in PMI Sweden, Chapter West and holds several certificates in Artificial Intelligence. In 2021, Marly together with then associate Jeanette Mill developed and launched an AI online educational suite for small and medium sized companies. 

Profile on LinkedIn

 

2022-03-09

Gå till toppen

Geetha Gopal

Head of Infrastructure Projects Delivery and Digital Transformation, Panasonic Asia Pacific, Singapore

Profile on LinkedIn

 
Geetha is a senior programme/project management professional with 14 years of experience leading large, strategic investments. In recognition of her contribution to the project management profession and IT innovation, she has been honoured by PMI in the prestigious Future 50 Leaders 2020 Global List.

 

 

Bruno Rafael Santos

Knowledge Management Volunteer / Community of Practice Initiative Coordinator

Profile on LinkedIn

 
Bruno Santos is a Knowledge Management Volunteer in PMI Rio de Janeiro Chapter and Project Manager at Coppetec Foundation, Federal University of Rio de Janeiro (UFRJ).He is leading the Community of Practice Initiative in the PMI Rio de Janeiro chapter. In UFRJ, he has managed Oil and Gas R&D projects since 2012.

 

5 Implications of Artificial Intelligence for Project Management

See article 

Artificial intelligence (AI) is no longer relegated to advanced IT and technology applications. AI has already begun making its way into more traditional settings to enhance or supplant tasks traditionally done by humans — including project managers.

What does that mean? What do project managers need to know to stay on the cutting edge of this emerging technology that has vast potential to impact the future of the profession?
 

1. Organizations — and project managers — see an imminent impact from AI.
Over 80% of respondents to a recent PMI “Pulse of the Profession®” survey report that their organizations are seeing an impact from AI. Over the next three years, project professionals expect the proportion of projects they manage using AI to jump from 23% to 37%, according to PMI’s “AI Innovators: Cracking the Code on Project Performance.”
 

2. AI-powered tools will take over administrative tasks for project managers.
AI-based tools can take over functions like meeting planning, reminders, day-to-day updates and other administrative tasks. This will free up project managers and team members to focus on higher-level, complex activities and planning.
 
According to a report from KPMG, “AI Transforming the Enterprise,” organizations who have invested in AI say they’ve seen, on average, a 15% improvement in productivity. In PMI’s “AI@Work” report, project leaders who are at the leading edge of AI and other technology frequently report that use of AI has cut the time they spend on activities like monitoring progress, managing documentation, and activity and resource planning.
 

3. AI systems can help keep projects on schedule and on budget.
“We’ve not been terribly good at estimating how much projects will cost and how long they'll take,” said Tom Davenport, professor and author of ‘The AI Advantage: How to Put the Artificial Intelligence Revolution to Work.’ By using AI-powered data analysis that looks at data from past projects, we'll be able to predict, with a much higher degree of confidence, how much a project will cost and how long it will take,” he explained on an episode of the PMI Center Stage podcast.
 

4. AI tools can analyze data from current and previous projects to provide insights.
Data analysis can do much more than estimate costs and schedules, however. Project managers will continue to steer projects through difficult decisions and unexpected obstacles, using AI for guidance and insights based on data.
 
“The real value of AI is in the algorithms that support decision making,” said Mohamed Hassan during a recent PMI webinar on AI Advances and Ethics. “AI will be an assistant to the project leader.”
 
Wanda Curlee echoed that sentiment in the webinar. “AI can bring forward important lessons learned to help project managers address risks and deliver better results,” she said.
 

5. Project managers will increasingly be called upon to implement AI-focused projects.
Rolling out AI implementations will require project managers to lead those projects. Does that mean project managers need to be AI experts? Not necessarily, but they do need to understand that these aren’t typical IT projects.