SKEMA BUSINESS SCHOOL MASTER THESIS PROJECT AND PROGRAMME MANAGEMENT AND BUSINESS DEVELOPMENT An Assessment of the impact of Machine Learning on Project Management By Victor Labrousse and Benjamin Bouvier Skema Business School 2017-2018 KEYWORDS AND ABSTRACT Year
SKEMA BUSINESS SCHOOL
PROJECT AND PROGRAMME MANAGEMENT AND BUSINESS DEVELOPMENT
An Assessment of the impact of Machine Learning on Project Management
By Victor Labrousse and Benjamin Bouvier
Skema Business School
KEYWORDS AND ABSTRACT
FIRST NAME: Victor
FAMILY NAME: Labrousse NATIONALITY: French
FIRST NAME: Benjamin
FAMILY NAME: Bouvier NATIONALITY: French
Artificial Intelligence / Project Management / Agile / Machine Learning / Consulting / Information Technology / Information Systems / Software Development
Amine EZZEROUALI, Director, MSc International HR and Performance Management, SKEMA Business School, Lille, France
ABSTRACTTo do last, 150 words, sum up of entire thesis
This Master Thesis has been written as of a graduation requirement for the Master of Science Project and Program Management and Business Development, under the supervision of Mr. Amine EZZEROUALI.
The chosen topic has been quite obvious for us as we were found of new technologies and very curious about Artificial Intelligence (AI). Studying into details this broad topic in relation to our field, Project Management, was an opportunity for us not only to gain understanding of a challenging and complex technology, Machine learning, but also to have a recognize expertise in both Artificial Intelligence and Project Management that could grant us a competitive advantage on the job market. Both of us have a different background and field experiences that allow us to have a broader understanding of both information technology on the one hand but also Project Management Consultancy on the other hand. The results of our analysis and researches have provided us with a clear understanding of how AI will revolutionize not only our field of expertize but the entire job market and our ways of working.
This thesis is not only the results of our own efforts. It has been a complete team work and there are some people without whom we would not have been able to complete this assignment. Thus, there are some special thanks we need to address.
First of all, we would like to thanks our supervisor Amine Ezzerouali for is constant help and devotion. This was not an easy task as few things were written on our topic and he helped us from day 1 to find relevant literature and provided us with his personal expertise in this particular field. He also gave us precious advises and all the way through. This paper would not have been possible without him and for that we deeply thank him.
We would also like to give special thanks to our MSc Director, Dr. Paul Gardiner, for his availability and friendliness during the entire time of our Master. Not only was he always available to help us on several issues and problems, he was also very understanding of our concerns and keen on finding a solution that would satisfy everyone. He was very comprehensive and this year has been nothing but perfect in many ways.
Special thanks also needs to be addressed to the Consultants of PCUBED, MI-GSO and the students from Skema, for their time and help in the interview and data gathering process. It is fair to say that without their help, we would not have been able to meet the quality standards we were aiming for. Thanks guys, we owe you one.
Finally we would like to thanks our friends and family for their tips and supports throughout our final years and especially during the draft of this thesis. Their emotional help cannot be quantifiable, and we could not have done it without them.
TABLE DES MATIERES
TOC o “1-3” h z u KEYWORDS AND ABSTRACT PAGEREF _Toc522611078 h 2PREFACE PAGEREF _Toc522611079 h 3EXECUTIVE SUMMARY PAGEREF _Toc522611080 h 6List of tables and figures PAGEREF _Toc522611081 h 7BODY OF THE TEXT PAGEREF _Toc522611082 h 81.iNTRODUCTION PAGEREF _Toc522611083 h 82.lITERATURE REVIEW PAGEREF _Toc522611084 h 102.1.Project Management Methodologies PAGEREF _Toc522611085 h 102.1.1The arrival of Agile PAGEREF _Toc522611086 h 102.1.2Project Management Future Challenges PAGEREF _Toc522611087 h 112.1.3From managing the change to changing management PAGEREF _Toc522611088 h 122.2.Artificial Intelligence PAGEREF _Toc522611089 h 142.2.1.AI: the game changer of PM jobs PAGEREF _Toc522611090 h 142.2.2.AI or the “Holy Grail” of accuracy PAGEREF _Toc522611091 h 153.mETHODOLOGy PAGEREF _Toc522611092 h 163.1.Introduction PAGEREF _Toc522611093 h 163.1.Research Strategy PAGEREF _Toc522611094 h 163.2.Research Methods PAGEREF _Toc522611095 h 163.3.Research Approach PAGEREF _Toc522611096 h 173.4.Data Collection Methods PAGEREF _Toc522611097 h 183.5.Sample Selection PAGEREF _Toc522611098 h 183.6.Research Process PAGEREF _Toc522611099 h 193.7.Data Analysis PAGEREF _Toc522611100 h 193.8.Limitations PAGEREF _Toc522611101 h 204.RESULTS PAGEREF _Toc522611102 h 214.1.Results from the survey PAGEREF _Toc522611103 h 214.2.Results from the interview guide PAGEREF _Toc522611104 h 275.dISCUSSION AND IMPLICATION PAGEREF _Toc522611105 h 34rEFERENCES PAGEREF _Toc522611106 h 34
“After writing up the final draft you need to write an executive summary which should be placed up front, just after the table of contents. In this summary you have to summarize the whole report; problem, method and findings. This summary should not exceed 3 pages.”
List of tables and figures TOC h z “Énumération 1” c Table 1: Features of Qualitative & Quantitative Research PAGEREF _Toc523737988 h 17Table 2: Overview of our Data gathering process PAGEREF _Toc523737989 h 19Figure 1: Age of respondent for the survey PAGEREF _Toc523737990 h 23Figure 2: Job Title of the respondents PAGEREF _Toc523737991 h 23Figure 3: Field of respondents PAGEREF _Toc523737992 h 24Figure 4: AI perception of the workers in the finance field PAGEREF _Toc523737993 h 24Figure 5: AI perception of respondents PAGEREF _Toc523737994 h 25Figure 6: AI knowledge of respondents PAGEREF _Toc523737995 h 26Figure 7 and 8: Impact rating of AI on respondent’s fields PAGEREF _Toc523737996 h 27Figure 9: The impact of AI on specific fields PAGEREF _Toc523737997 h 28Figure 10: The development of AI; opportunity or threat ? PAGEREF _Toc523737998 h 31
BODY OF THE TEXT
iNTRODUCTIONProject management has been around for centuries. Some are saying that the construction of the great pyramids of Giza is teaching us about how projects were managed at the time. But the profession of Project Manager, the one with the frameworks and the standards we have nowadays is a rather new profession. It is evolving incredibly fast, thanks to breakthrough in technology, software, and social theories. A great amount of tools and methodologies have risen up through the last decade, revolutionizing the way we use to handle projects. Project Manager needs to be fast thinkers, ready to adapt to new opportunities, and they are facing more and more changes in customer demands and expectations. That was one of the main reasons for the Agile methodology to come on the market. Customers want to be able to handle change rapidly, with a faster time-to-market and more flexible requirements. Agile help address those current needs, but we should not neglect to plan for the future needs the market will have, and how we will have to address them.
One of the greatest technological challenge of our time is related to Artificial Intelligence (AI). AI is not new. It has become an academic discipline in 1956, and is evolving at a very fast pace. AI is “everything that is not done yet” by machines and computer technology, in the sense that when something has become a routine task, it is not considered as AI any more. This was the case for facial recognition by machines. AI is focusing on two key areas: problem solving and Machine Learning. Machine learning is a field in computer science that allow a machine to “learn” and give predictions, without being explicitly programmed. Machine learning is not new, but like artificial intelligence, it is spreading and developing at an incredible pace in the last ten years. Ai development may disrupt a lot of things in our current ways of working, in the way of a new industrial revolution. Hence, as future Project Managers, it seemed important to us to try and seek what the impact will be and try to be ready for the change. This thesis is to be used as a guide to understand the impacts AI might have on our profession.
We think that Machine learning offers a lot of possibility for Project management, especially when it comes to machines that are able to develop their own “skills” and even end up having a greater knowledge than the person who programmed it. How are those new technologies going to bewilder the profession as we know it? What are the new skills Project Manager will have to have in order to use those new tools, and these technological resources? Will it become the management of computers more than the management of the people? The aim of this master thesis is to provide a guide on how AI will change the PM profession, and what could be done in order to adapt to and embrace this change.
Our assumption is that Agile will not be able to cope with such a change. Eventually, we will need a new way of working, new methods. This can mean a new methodology, or a redefinition of our profession. We make the hypothesis that AI, and especially Machine Learning, will have a strong impact on many Project Management tasks, such as risk management, resourcing, training, and will redesign the profession as we know it. Not only that, we do think that in order to prepare for such a challenge for our profession, the most efficient thing to do is to prepare for the arrival of AI, by trying to predict how new predictive machines will disrupt our daily working life.
Our First focus will be to demonstrate the complexity of trying to predict a future that is uncertain. Secondly, we will research and explain the potential impacts of Artificial Intelligence and Machine Learning on important areas linked to Project Management and on the peoples needed to carry out Project tasks. Finally, we will try to give recommendations and advices on how, regarding that future, we can hope to adapt ourselves and our services, and how to be ready for the upcoming changes. This has been done by conducting important research using data gathering tools, that we had to analyse in order to gather qualitative and quantitative data that give us an insight on Project Manager’s experience and opinions towards this disruption that we will face.
lITERATURE REVIEWThe literature review has been conducted through the careful study of papers, books,
Project Management journals, and articles. This last sort of source has been very important for us, as few has been written on the impact of AI on Project Management. We struggled to find evidences and previous researches on the topic. Hence, we decided to use professional articles, supported by trusted journals and sources, to guide us through the literature review. The use of articles being different, it cannot be trusted without a thorough analysis, to avoid limitations and misleading data.
The literature review is composed of two main topics, hence this is not a chronological approach. The topics are the following:
Project Management Methodologies
Project Management MethodologiesThe arrival of Agile
Project management has faced many changes over the last decades. We saw the arrival of a new approach based on the Agile Manifesto in 2001. This book revolutionized the way we use to approach and handle projects, and several tools have now been developed to help companies and Project Managers to be more “Agile”. The traditional approach in Project Management was the Waterfall model. Waterfall is an approach where every phase of the project are subsequent, with a finish-to-start relationship. The project will flow from one end to the other regularly, such as a waterfall. This model was inherited by enterprise software developers (Bowes, 2014). As Bowes explains, Waterfall has been around for decades. And many issues were raised using this model. It was seen as an “inflexible” model that was unable to “respond on aggressive customer requests” and thus could not take change into account (Papadopoulos, 2014).
Papadopoulos has been a fervent defender of the need for an Agile mind-set in nowadays industry. He describes Agile as a set of methodologies and good practices that enables companies to embrace and respond faster to the fast-changing world we are living in. Project Management faced a challenge as its over-used traditional approach was flawed, and this was mainly discovered thanks to new technologies (Bow, 2014). When using waterfall in Software development, the fixing of the scope and design phase were done before the beginning of the coding. The developers are often not able to forecast the problems that they may face and thus cannot adapt to or mitigate those. The planning was hence not accurate. Besides, the requirements were fixed. This means that any change in requirements will put the product back up on the waterfall at the beginning of the process, for the design phase to start over, thus causing major delays and potential costs increase. In a perfect world the requirements would not change, but more and more clients are looking towards flexible solutions, as often, the developers know more about what they can do and when then the customer of the managers. Agile is all the more interesting for companies has it can be tailored to suits any kind of Projects, and not just used for software development.
The findings of Papadopoulos tend to show empirically that Agile projects, even large Agile projects, (as it has been a critics towards Agile that it could not be implemented for Large Projects) are better performing then more traditional one. This has been seen on the customer relationships, on the end product quality, and a better collaboration within the team.
Project Management Future ChallengesAgile has changed the PM profession as we know it. But this paper needs to be focussing on whether the future of Project Management can be predicted, especially regarding the new technologies that will arise on the market. One way of doing it would be by looking at the keywords searched within Project Management journals (Uchitpe et al., 2016). They propose a new approach they called “Relative Growth” which aims at looking the most searched keywords in comparison with the ones having the highest growth in terms of number of occurrence.
Their findings indicates that among the emerging keywords, several can be linked to the agile mind-set such as Portfolio, Enable, Positive, Culture and Learn. This indicates a trend but cannot be taken as an empirical proof of where the profession is headed. In fact, the weakness of their analysis and research is that they based the keyword search on only one journal: the International Journal of Project Management, IJPM). This is a “strong foundation” for further research, as they conclude. Another approach for trying to identify how technology may affect Project Management would be the one adopted in the report conducted by the Management Consultancy Agency (MCA) in The definitive guide to UK Consulting Industry Statistics 2017 Consulting in the Age of Disruption. Their approach was based upon the thought that no one knows better how the Project Management profession is heading to but professionals (MCA, 2017). This recent report aimed at interviewing consultants in Project Management about their thinking of the changes and challenges they will face in the future. Again, nothing should be taken for granted as per the fact that they are giving their opinion based on few statistics. What is useful for our study is the fact that most of the consultants are having the same idea of what will happen next. Even by having different backgrounds, different companies and field of expertise. This is also showing their current state of mind about the technological disruption.
The conclusions that can be drawn upon the interviews is that Project Management is facing a rather new and huge challenge that is continuously fast-growing: the technological disruption. Consultants within the MCA report states that Project Management has to deal with the fact that disruptive technological technologies are accelerating the pace of the change. As a result, the Project Manager needs to be aware that their role is likely to change, and they will have to adapt to this new digitally connected world (MCA, 2017). We are facing smarter industries, a blend of automation and Human resources. These are the changes on the shorter term. On the longer term, due to the complexity of the Digitalisation, things are harder to predict. And according to Bowes’ description of Agile, it mostly relates to being more flexible and adaptive, and a key component of its methodology is linked to responding rapidly to change (Bowes, 2014).
Agile could be the key to addressing those challenges on the short term. The limits of the MCA documents is that these forecasts tends to show a trend in the digitalisation, but this is a broad term, where AI alongside with many others technological tools can fit. It is far from being accurate, and there is no empirical proofs that shows how Agility could help address the potential issues faced by Project Management, especially when it comes to Artificial Intelligence.
From managing the change to changing management
We have seen that due to the complexity of the digitalisation, the long term forecast of the change Project Management will go through is hard to predict. Besides, one of the limits of the MCA report is that it does not specifically talks about where AI would fit in this future. But the MIT Sloan Management review have explored a different approach. Their researches are based upon fifteen articles written by fifteen experts. They were all asked to answer the same question: “Within the next five years, how will technology change the practice of management in a way we have not yet witnessed?” This aims at having a better understanding on how technologies will affect management (MIT Sloan, 2016).
This set or articles gives insights on several topics and areas of focus, several points of view of people experts in their own fields. What is interesting is comparing the expert’s opinions. Regarding how management will evolve, the conclusion are that the role is likely to go from a moment where we manage people to a moment where we would manage the bots that are managing the business. Workers are now in some companies the software build by humans. To some extent, those software developers are their managers. Through feedback, they evaluate and “coach” their workers through updates (O’reilly, 2016). The managers won’t become obsolete but they will need to have more technical skills than what is needed nowadays. O’reilly gives provides a good analysis alongside with a case study of how Uber is dominating the market thanks to its AI, and to some extent, how Uber as become both manager and managed by their data software. But it tends to be really AI friendly in the sense that the potential grey areas of AI are not mentioned in his article.
Regarding management and ethics, the future brought by AI tends to be more mitigated. It is a fact expressed in most articles that advances in information systems have been highly beneficial in our everyday lives. But there could be downsides (Parmar and Edward Freeman, 2016). In their article, unlike others articles from the journal, the authors discuss the ethics that lies behind the algorithms, as they are made by a human. And humans make mistakes as we are not perfect and full of judgments. This represents a real life problem that needs to be addressed. They intent to make readers aware that when it comes to ethics, we are not sure that we want machines to make our decision for us. This problems has been enlightened by the challenges software developers faced when designing self-driving cars. If it comes to this, would we have the car prioritize the life of people within or outside the car? This is a thorny issue that will need to be addressed in the future. Project managers will need to consider this as going to be a continually increasing part of their everyday job life.
One of the biggest management challenge will be to understand how to get the most out of both humans and machines (O’reilly, 2016). AI will change the way we manage and are managed. We need to prepare for the future if we want to be able to make the best out of it.
AI: the game changer of PM jobs 37% of the work currently performed by UK workers are forecasted to be handled by machines within two decades (Carl Benedikt Frey and Michael A. Osborne, 2013). Behind the general word “machines” not only physical tasks are concerned but also intellectual ones and that in every fields. Project management is no exception to this trend and is expecting to gain a lot more efficiency from a branch of computerization which is Artificial intelligence. The enthusiasm is so striking since investment in AI are projected to reach 3 billion USD in 2024, compared to 126 billion in 2015 according to a 2016 Transparency Market Research report.
We are now experiencing a new era: the one of digitalization and artificial intelligence. But what are those two concepts? In the common mind digitalization is appears as a dematerialization of the economy and services. Artificial intelligence seems both an empowerment of people which would develop their capacity and productivity, a possible threat to their job considering the limits of the human efficiency and even a threat to their safety and their power over their own environment. The truth is that the digitalization on one hand, is the increase of the use of digitization (the action of converting images, sounds or texts into data computers can process) by an organization, industry, country,… Artificial intelligence on the other hand “covers everything done by a computer or machine that resembles human thought” (Hosley, W. N. (1987). The application of artificial intelligence software to project management. Project Management Journal, 18(3), 73–75.
But the idea that AI would really destroy employment in Project Management it is not shared. Indeed, it would be better to imagine a change in the work managed by PM officers. Through the implementation of AI via software, project management workers will save a huge time that they could dedicate on high-value functions (Boris Petukhov, n.d.). Then instead of a loss of jobs it would be better to imagine a change in the scope of work of project management officers who would focus their work on high valued tasks. Also what should be taken into consideration is structure of employment in terms of age and gender. Without human biased people would only be hired in terms of skills.
AI or the “Holy Grail” of accuracy”Machine learning can estimate effort and cost for each work breakdown structure with better accuracy, using the input of details like type of activity, resources involved, environment and skill set required, along with historical project data,” (Saravanan Mugund, n.d.). If there is one notion that he is highly spreading from pro AI movement, it is the idea that AI enables the collection and the process of a huge quantity of data through data mining. The impartial process of such an amount of data appears impossible for people but is realizable through expert systems. There the different data integrated in the software are converted into a tree of decisions who helps in decision making and what action plan to choose. It helps in different areas of Project management such as planning, managing risks under uncertainty designing quality assurance.
However what is not taken into account is the fact that AI and computers do what they’re programmed to do. They are designed by humans and they are following a human, possibly biased, way of thinking. Machine learning is still not the smart and auto-deterministic process Ada Lovelace was expecting. Also before experts systems are giving you different scenarios, variables need to be manually introduced and weighted by a Project Management officer. In this case flaws remains even with the help of AI.
mETHODOLOGy IntroductionAs stated in the title, we will be describing in the following part the research process we went through and our methodology towards data analysis. We will hence go through our research strategy, methods and approach. The sample selection process will also be described, alongside with our data analysis methodology. We will conclude the chapter by explaining the limitations that need to be considered when reading our paper and the results of our data gathering journey.
Research StrategyWe struggled to find relevant data and existing literature on our topic. This topic is not brand new but when applied specifically to Project Management, relevant literature become a scarce resource. We hence broaden our research on relevant papers articles and journals. As such, we based our research on both new and existing researches.
Research MethodsIn order to comply with what we wanted to achieve through our study, we decided to go for a Pragmatic approach. Using a Pragmatic approach has been obvious for us as we needed to use both quantitative and qualitative approaches to measure respondent’s point of view on AI, and on its impact on Project Management. Both have they perks and we wanted to get as many useful data as we could, without getting caught in one unique approach. Hence we conducted semi structured interviews with professionals from either Project Management background or Information Technology experiences. We also constructed a survey designed to be answered by a larger scaled sample, so we could extract empirical qualitative statistics. Gathering qualitative data are not an easy task as they are often biased by the researcher’s own interpretations of the answers. We do think that the fact we are two co-authors of this thesis helps reduce the bias and interpretations to a minimum acceptable level.
Table 1: Features of Qualitative ; Quantitative ResearchQualitative research Quantitative research Pragmatic research
We aim for a complete and detailed description
Classify and count features and variables to build statistical models Use of any methods that is the best suited for the research topic
Researchers only have a general idea of what they are looking for Researchers knows clearly what they are looking for Researchers draw hypothesis and try to either confirm or refute them
Recommended earlier in the process, design emerge as study unfolds Recommended later, design fixed way before data is collected Can be used at any time when needed, design is set when it needs to be
Researchers is the data gathering tool Researchers is using tools such as surveys to gather numerical data Researcher can use anything relevant according to the data he want to collects
Researcher is more subjective Researcher is more Objective Researcher can be both objective and subjective according to the chosen method
Data collected are sentences, words and picture Data are numbers and statistics and numbers Data can be in any format
Adapted from: Miles ; Huberman (1994, p. 40). Qualitative Data Analysis, available
Research ApproachFor the purpose of our paper, we decided to use the both the Inductive and the Deductive approaches. We started with hypothesis based on observation and experiences that are drafted toward the end of the research process. The hypothesis we made at the beginning of the research process where not set in stone, and we kept the freedom to change and alter them according to our findings during the duration of the research process, meaning that our approach was not entirely Top down nor Bottom up. With our research topic, we needed both open-ended questions and pre-specified ones. Our research method being pragmatic, we knew we wanted to analyse both numerical and descriptive data, with different sample size. One big for the survey and one small for the interview guide. Both methods have their pros and cons hence we need to be careful of not falling into both methods downsides, such as objectivity issues and too small sized samples.
Data Collection Methods
We decided to use two types of data gathering methodologies. The first one was the conduction of semi structured interviews. We drafted an interview guide to for our semi structured interviews but we would not just stick to the questions written. We took the liberty to go off track sometimes, asking interviewees to develop or explain several points that seemed relevant or important for the sake of our research. Besides, as this is the most flexible methodology, this matched our pragmatic research approach. Unlike unstructured interview, the conversation will remains under control while there is still room for an informal chat that can provide us with insights on theories and results we might not have considered at the beginning of the research process.
For the sample not to be too small and to broaden the scope of the interviewees, we also decided to draft a survey with structured fixed questions to gather a great amount of data targeting to refute or prove the conclusion made with the interview guide. As a result this was a structured survey, used to gather the proper information we needed to further comment what we already collected, and see on key aspects whether a trend could be observed when going to a broaden sample.
Table 2: Overview of our Data gathering processSurvey Interview Guide
Number of interviewees 49 14
Number of questions 10 17
To include example of questions and survey
Sample SelectionFor the purpose of our research, we decided to use two types of sampling. For our interview guide, which was targeted a smaller sample, we used the Purposive sampling methodology. We had our criteria, and needed people that would fit either one or the other. Our criteria were either to be very familiar with IT and IS technology, with an enthusiast towards AI and new technologies, or to have a deep understanding of Project Management, and field experience. Although this selection method is not representative of the population, we needed skilled people to answer specific question, whose answer might either confirm our theories, or broaden our understanding of our chosen topic. That is why the respondent of the interview guide were either Senior Project Management Consultant, or IT Specialist, sometimes both.
As of the survey, which was designed for a larger sample, we used both the Convenience Sampling coupled with the Purposive sampling. In fact, we needed respondent with a basic to good understanding of AI and/or Project Management. As of the convenience, to gather many relevant responses in a short amount of time, we used the social group from Skema, with former students. Using this method we managed to gather plenty of relevant data in a short amount of time. As a result of our sampling choices, we interviewed and gathered answers from 4 main sources:
PCUBED Senior consultants and executives
IT-Specialist from …..
Current and former SKEMA students
The surveys were launched/sent in July 2018 and ran through the beginning of August. The interviews with executives and consultants were conducted in August, most of them over the phone as per convenience; we could not always be where the respondents were. Others were sent through emails, to be answered when the consultant had the time, followed by a discussion when needed to discuss further the answers. The interviews went well and smoothly and lasted less than an hour. We kept notes and records of what was being said as part of the interview, for data gathering purpose.
As said in the paragraph 3.5, we conducted semi structured interviews. This meant that even though we were following an interview guide we sometimes took the liberty to ask the respondent to develop what was being said, and asked extra questions. The interview process went great and the discussion with the respondents were very pleasant.
Data AnalysisIn order to analyse the data we gathered, we chose to use content analysis. We categorized the data we collected in order to compare them. We also used the answers we got to pull out statistics among the respondent to speculate on a more global impact. We compared the viewpoint of both professionals in Project Management and IT industry, alongside with the viewpoint of business students with an IT background knowledge. This gave us quality quantitative results, using a qualitative interview guides. Based on our data and analysis we draw conclusions that will be discussed in the next part of our research paper.
We are aware that our research process is not perfect and hence we must share the following limitations to our study:
As per the interview guide, the size of the sample is rather small. Meaning that it cannot be very accurately representative of the entire population. Especially when considering that this is not a homogeneous sample of the population as we selected it. This, even though in line with our topic, needs to be considered when reading this paper.
Analysing and measuring qualitative data is not always accurate. The respondent and the interviewers are humans and hence subjects to mistakes. They may feel unsafe to say things regarding their position within the company, or say things that they unconsciously think we want to hear, to make us satisfy. Even though we took that into account this is a limitation of our study.
Qualitative data were subject to interpretations by us, meaning that we always run the risk of misinterpret the data collected. We are never perfectly objective, and our bias could be a limit to the study, despite the fact that two authors can challenge each other points of view and reduce the potential bias.
RESULTSIn this section, we introduce our research results, gathered from both the survey we conducted and the interview conducted using the interview guide as a guidance. The results are important for either proving or refuting our hypothesis, and explore correlation we did not have in mind at first. As explained in the third chapter, we used different methodologies and tools to help us during the data collection process, thus giving us insights on the way people feel towards AI, and a deeper explanation made by professionals on its potential impact. The results are to be put back in their context for a better understanding of both what we found and the limitations resulting from our chosen methodologies. As we used two main tools, a survey and a interview guide, the results will be disclosed accordingly in two different part. Later on in the next part, links will be made between the data gathered from the survey and the interview guide to better show correlation and differences among the data.
Results from the surveyIn the third chapter, we explained that we used the Convenience Sampling to gather data for the survey. In fact, we had access to a social network linked with our school that allowed us to collect a good amount of data in a rather short amount of time. That platform allowed us to gain time and effort. We also wanted to keep the age variable close to ours to look for a different opinion and view that would not be biased by experience. We wanted to know whether they would have a different point of view, and most important the reasons for a divergence in ideas. Hence, 69% of our respondent are aged between 20 and 25. We were faced with an issue, with respondents not being willing to provide their age. We do know that they are less than 40.
Figure 1: Age of respondent for the survey
Another variable that was worth knowing was the work experience of the respondent. This data link with the field in which they are working / studying provide us with intel on how knowledgeable they are, and how familiar they may be on either Project Management and AI. Our hypothesis is that AI perception can be explained at least partly by looking at people’s background and expertise/field. As per the age of our respondents, 85% are either Students or Interns.
Figure 2: Job Title of the respondents
Part of our hypothesis is that there is a correlation between people’s fields and their perception on AI. Thus we asked respondents to provide us with their field, so that we could link it with their job title, and their own AI perception.
Figure 3: Field of respondents
It is important to emphasize that the existence of such a link seem to exist, but does not appear to be automatic for every field. A strong 39% of our respondents come from a Project Management background, which is explained partly by our convenience survey and by our intention to have people answering the survey knowledgeable of Project Management, such as us, as this is one important dimension of our chosen topic. 11% of our respondents are working in the Finance department, where the AI debate is omnipresent. This will be very important for our analysis. In fact this is one of the data that allow us to think that there is indeed a link between the field and the perception of AI.
Figure 4: AI perception of the workers in the finance field
Indeed, 80% of the respondents in the Finance field think of AI as an opportunity, with 20% thinking that it can also be seen as a challenge. This is specific to the finance field. When looking at the overall perception the data gathered are slightly different. This will be detailed in the following chapter.
For our research topic, we needed to know how AI was perceived by our respondent. Such as for the Finance department, we needed to see whether this could be linked to any relevant data that could explain such a perception. The following figure shows what people think of AI.
For a staggering 41% of people, AI represent an opportunity.
Figure 5: AI perception of respondents
Looking closer at the data, it appears that for 85% of people, AI is either an opportunity needed in today’s business environment, or a challenge that needs to be addressed in order to be used properly and provide benefits. When asking the respondent what motivated their perception, people’s answers tends to be quite similar; for those thinking that this is an opportunity, they often say that “the predictive behaviour of AI will truly help one’s business to flourish”. Likewise, those who thinks that this is a challenge say the same, while arguing that without being handle properly, may end up having less skilled workers lose their job and that ethical concerns will be raised. For those thinking that AI is a threat, they mostly justify it with job losses, or even consider the dystopia where “machines will at some point take over humans”. It appears important here to emphasize that there appear to be a correlation between the facts that those who argue that AI only represent a threat are not among the most knowledgeable on AI.
Figure 6: AI knowledge of respondents
When asked to assess their knowledge on AI, respondents tend to be quite pessimists. Indeed, 59% of our respondents do think that their own personal knowledge on AI is less than average. This is what we forecasted when selecting our topic. Due to the fact that AI is a massively broad topic, peoples tend to be confused and assume that they don’t know much about it. We still have 42% of respondents with a better understanding of it, including 2% that self-qualify them as expert. As said previously, those 2% are mainly working in the Finance field where the AI discussion is very important and present.
Figure 7 and 8: Impact rating of AI on respondent’s fields
The last two graphs shows how strong respondents rate the impact of AI on their field, and on their day to day activities. These last two graphs are critical to understand the state of mind people have towards AI. We saw earlier that for a majority of people, AI will be an opportunity if handle properly. Some still see it only as a threat, fearing a never-seen-before dystopia. But whether positive or negative, people mainly do agree that its impact on our day-to-day activities will be very important. More than 74% of the respondents think that on a scale from one to ten, the impact will be rated 7 or above. For every question we asked, people’s opinion tend to be distinctive, either seeing AI as a positive thing, or a negative one. But on this one, the opinion tend to converge. According to the data we gathered from the survey, the question shift from is AI will disrupt our lives, to how will it change the way we work.
Figure 9: The impact of AI on specific fields
When asked what will be the most impacted, we can also observe a convergence in point of view: in fact, people massively think that Data analysis will be what will be the most disrupted by the development of AI. On the opposite, out of 142 answers, Team Management only come out 2 times. No matter what, there seem to be a correlation between the AI disruption and Data Analysis. This seem to confirm one of our hypothesis. The predictive potential of AI will disrupt Data analysis, which seem to explain the fact that the finance department will be one of the most affected by the development of Artificial Intelligence. The impact will more likely spread upon several fields but they won’t be impacted all as much. Such a statement need to be analysed more deeply in the following chapter. What confirmed our main hypothesis is the fact that out of our respondents, no one think that AI won’t change anything in either their field or their day to day activities. They may have rated the change as low (3/10) but never below, which tends to indicate that everyone recognize in AI its disruptive power.
Results from the interview guide
Now that we have reported the findings from the answers gathered with our survey, we need to report data from the interview guide. Unlike the survey, for the purpose of a deeper analysis, we used personal and professional experience from professionals, mainly Consultants with a Project Management background. The answers are less numerous than the survey, considering the fact that we conducted interviews, asking more specific questions to have a better understanding of the potential impact of AI . Within the interviewed company, the AI topic is gathering pace and is getting important. There are ongoing discussions to how we can organize internally the company to better face those new challenges that are coming our way. Not only are the interviewees more likely to be more knowledgeable on AI, but they also give us a good and precise insight on its disruptive potential and how one can try to adapt in order to be as ready as possible to tackle new thorny issues.
The age range is already more representative of nowadays’ business environment, with people aged between 21 and 58. Most of them are consultants, but their seniority vary from Junior to Senior consultant, and C-level executives. Their understanding of Artificial Intelligence is quite developed. This is seen when they are asked to redefine AI according their understanding. We can see a pattern in the answers with a lot of similarities. As said previously, AI is hard to define as per the fact that this is a very broad topic, and as it encompass several domains. But they managed to give us precise definitions, by redefining the context. One consultant defines it as being “Cool things that computers can’t do yet” but he also advises us to limit the discussion to neural network and machine learning when considering the professional applications. This definition may seem funny at first but this is very accurate. One the one hand the definition is quite exact, as Artificial Intelligence is everything that computers cannot do. Once it can be done by a computer, this is no longer considered as Artificial Intelligence. The next part of his definition is proving us that despite the funny ton of his voice, he truly knows what he is talking about. This particular answer is not only accurate, but also significant of how well informed our interviewees are on AI. The data they will provide us are to be put back in their context with their own limitations, but will be very useful for our analysis.
For a majority of them, the AI debate is not omnipresent in nowadays’ society. It is indeed present in some specific industries, where AI has been mainly seen as an opportunity, but in other mainstream discussion, where the AI should be present to ensure a smooth transition, it is not present enough. For one consultant, the mainstream machine learning debate has been “more about driverless Uber rather than what will be more important for companies such as decision making in factories based on environmental factors”. That would mean that when we hear the word AI, this is misused and the debate will “gather momentum as machine learning will revolutionize the industries” and that with its potential, the debate should be more present. Other consultants have a mitigated point of view, saying that yes, it has been present where highly relevant, but not enough in industries where the impact has not been identified yet, which confirm the burning issue we are trying to answer in this paper. What will be the impact of Machine Learning? It is mainly agreed within the consultancy environment that the mainstream debate is misunderstood and wrong, and that for companies and industries thinking that machine learning won’t have such an impact, the disruption will be even more important. The debate might just help companies prepare for the change coming their way and not having a serious discussion on it might jeopardize the industries not prepared for it, which confirm our hypothesis. One consultant is particularly negative, stating that “the advent of the 4th industrial revolution of which AI and associated technologies will profoundly change our economy and society, to the degree that current political and social structures may not survive.”
Consultants have a very similar opinion than our respondents from the survey, on whether AI will likely be an opportunity or a threat. A majority of them are optimistic, and do think that despite its disruptive potential, AI will more likely be an opportunity. That is the case for 60% of them. 10% of them still thinks that AI will at some point take over humans which seems linked to the general knowledge of people on what we call artificial intelligence. We do think this is caused by the misunderstanding towards the globally accepted definition of AI, and this needs to be discussed in the next chapter. For most of the consultants, AI does not have to be a threat. One even draw a parallel with another disruptive challenge our industries have faced in the last decades; the arrival of machineries in factories. For most people, this meant “machine taking people’s job and increase in unemployment”. But we still do need people in factories, it has just changed our way of working and the skills we needed in factories back then. “For AI it can be the same, it will not eliminate the need for people in the industries”.
Figure 10: The development of AI; opportunity or threat ?
For respectively 55% and 27% of the consultants, it has the potential to be an opportunity and both a threat and an opportunity. This was surprising as we made the hypothesis that consultants, with a deeper understanding of the job market and the Project Management industry, would have a less optimistic view, more mitigated. Those who think it can be both says that it has the power to compromise many roles, from low skilled to skilled ones. “There are many opportunities for companies with the right blend of skills to monetise hi-volume data” but that can be threatening for other companies which are not. As a tool, AI and Machine learning, depending on the user can be used for good and bad purposes. The most surprising answer was the consultants thinking that it wouldn’t be either good or bad, point of view that accounts for 9% of our data. This was the only answer we did not forecast before asking the questions. One of the consultants explains it as something that can, at the same time, increase threats, and/or enable opportunities.
When asked whether AI/Machine learning will impact or way of working, a surprising 100% of them answers yes, but this has to be mitigated. In fact, most of them are planning this impact to occur on the medium to longer term. One of them, being more precise, say we will start to see an impact on our way of working in 18months. Other see this change within 5 to 10 years. The timeline is not quite unanimous, which was something we forecasted in our literature review. Not only that, but according to their fields, they all see different potential impact on the medium to longer term. One explains that it will be at first the middle skilled job that are going to be impacted. That impact will, on the longer term, spread on the lower en higher skilled workforce. Another confirms our hypothesis that the finance field will be the most impacted: “In the finance world where I work we will see more hedge funds and trading run by AI and algorithms”. But he also sees over impacts such as the Admin tasks. To him, the admin functions are doomed to disappear, replaced by AI. They all see different potential impacts on the way we currently work. Getting into the topic we are more interested in, they also have very interesting forecasts on their own field; Project Management.
To them, Project Management as we currently know it will deeply change, and we need to be ready for it. When it comes to process, they think we will face an automation of the processes and the entire role of the PM will shift. But again, hypothetically due to the very unpredictable nature of the future, they point of view tend to diverge. Some think that it will have to be tailored to each project as what PM does is mostly unique and linked to one and only one project. He explains that “the side that involves less people, like tracking budgets could be done with an AI system” but the rest would have to be specific to the project on a case by case approach. This thought tend to be shared among the consultants, especially regarding the automation. One said “Manual processes covered by Project Manager’s responsibilities not relating to soft-skills/people management” will be one of the major shift. To put it simply, “it will manage for us” meaning that Project Managers are less likely to handle some of the tasks we are doing nowadays on a day to day basis. One consultant think that PM will more likely become coach for people who don’t do Project Management, which will be a throwback in true coaching/consulting activities. Several share the opinion that easy tasks will be automated such as planning and project updates. PM may have in a near future more time to allocate to team management and conflict resolution. What will change deeply for one of the consultant is Data gathering and analysis, alongside with reporting activities. He explains that “AI/ML is the trend toward better structured and more centralised data, and linked to intuitive report design tools. There will be less demand for admin overhead such as PMO headcount.” But to him this impact won’t be for every company, as most organisations don’t have enough data to allow for AI; only large ones with many customers, or industries that share data will. Some have a more negative opinion on the impact on Project Management, with the potential end of the PMO, replaced by AI, when other considered the possibility of Project improvement with data analysed by machines and warning signs / corrective actions taken sooner into account and identified / done faster than what a human can do. The PMO disappearance, as a consultant explains, has already started in some companies, and he mentions Airbus, with the Product Control Tower Initiative (PCT). We will discuss this project and its implications in the following part. It might not just be the disappearance of well-known roles and functions, but also the chance to create new positions, as one senior consultant explains: “New roles will form in designing information infrastructures for project management information but there will be fewer of them and more highly skilled”. The impacts will be plural and may also depend on the human worker’s skills. As a consultant explains, “It depends on the consultant’s aptitude, industry, inclination and project. What impacts me the most may not impact someone else at all.” Which tend to confirm our statement regarding the hardness to predict the impacts of AI on our lives, and work environment.
One of our hypothesis is that we might be able to tailor an existing Project Management methodology, Agile, to handle such a change. Agile is a rather new way of delivering project that has been commonly accepted as a better way to handle most of our project by focusing on an early delivery of a working solution that will be continuously improved overtime. Simplicity and efficiency are key when it comes to Agile. In the Agile manifesto, change in the project environment is meant to be embraced as it is a mean to reach a better product/output. Hence, our question was is Agile going to be agile enough to handle a massive change such as the arrival and development of Artificial Intelligence and Machine Learning in Project Management? Can we tailor it to best take advantage of it and reduce the risks behind it? Despite being Agile Certified, as most of our respondents in Consultancy are (73% are very familiar of Agile), we needed to ask this question to people having actually ran Agile project. We had a 100% of Yes answer for this thorny question, which was surprising. But among them, 27% are mitigated, and thinks that it will have to be modified in order to do so. Indeed, it appears that in its nature today, it might not be designed properly for such a change. Many consultants agree with that idea. By having the choice between Waterfall and Agile, Project Managers “won’t have much of a choice” as it is “better optimised than Waterfall” nowadays to run AI projects. But as the change might be incrementally bigger and deeper, Agile might “need to be reinvented a bit” as explains one consultant. Agile is perfect for software development, but made by humans, by software developer. If Ai at some point make the development for the developers, we will definitely need either a new methodology, or a more tailored Agile approach. What we do need to take into account when tailoring it is to be clear about “what information and decision we want from the humans and from the AI”, and especially how the AI will support the decision making.
The mentality we need in order to be ready for the change is mainly to be open to the change but keeping a bit of scepticism to where AI could go wrong and avoid bad surprises. Another really interesting point made by a consultant is one really important skill identified, which is reskilling, one’s ability to adapt and learn in order not to be “left behind” when the change will occur. We did not consider that particular skill but this will be very interesting to analyse in the next part when considering the how to prepare for such a change.
AI will not just be an opportunity to improve our Project delivery rate and planning accuracy. For 9% of our interviewed consultants, it mainly represent a threat. Indeed, Ai raises several concerns, especially when it comes to Cyber security and ethics. We saw a rise of such ethical issues when developing self-driving cars, and the thorny issues that was “do we value more the life of the people within the vehicle more than the lives of the bystanders if an accident is inevitable”. Most of our respondent do not think that we will see an improvement of ethics in companies thanks to AI. Our interviewees are afraid that job losses will arise because of AI, and that companies don’t “care that much about ethics, unless the market dictates so”. The ethical concerns seems to be correlated with the fact that people see AI as a threat. But some consultants, who do think that AI is more of an opportunity than a threat, actually think that we might see an improvement of the working conditions and that companies, because of the massive data exchange caused by AI, will have to be more open, more transparent, and that it might cause an ethical improvement of companies. But the other concerns with letting algorithm run our business is about data security and Cybersecurity. Indeed we rely on the data to run our business. Already important in our society, it is to become more and more important in the near to long term future according to our respondents. “There is a threat that the AI will be autonomous, and decides to compromise the security itself if it thinks that it is for the better”. Companies will have to be even more aware of hacking issues, and will need to know how to efficiently protect its data, by “encryption, IT integration and manual processes”. One consultants goes further, explaining that we will also be able to use another AI to enhance cyber security and hence protect our other AI related data. But so can hackers. “AI, as a tool, will be able to be used as a semi-independent hacking or counter hacking roles”. The problem might be for the humans to keep up with the AI tools as they rapidly learn & evolve.
dISCUSSION AND IMPLICATIONIn this section, we will be developing our findings presented in the result section. Using the insights gathered from the semi-structured interview the survey we conducted, we will be assessing and discussing the impact of machine learning on project management. In order to explain the correlation we found when looking at the data we gathered, we will also be comparing our findings to the existing literature gathered and analysed to answer this topic. This section will be divided into key ideas to facilitate the reading and the understanding of the argumentation.
The evolution of Project Management
Through the findings and the results of the survey, the possibility of a drastic change in PM practices has been highlighted by many interviewees. In that sense, every activities done by PM Officers seems to be impacted from day to day mechanical tasks to people management or recruitment. PMO’s as we know them are seen to be on its downfall, and it is important to understand the reasons behind the changes. We will discuss the magnitude of its impact on the angle on the profiling of the new project management officer.
Another typology of project management officers
“AI is not going to replace project managers. But it is definitely going to be a key tool in the project manager’s toolbox to improve delivery practices”, (Mugund, n.d). This is the position defended by the majority of the consultant we have interviewed (55 % of them). What remains is the necessity to involve Project management professionals, but their scope of action will be reduced. Monitoring a project will be easier and more detailed: the huge amount of data collected from the progress achieved are to be reported more frequently and in an understandable format. Project management officers will no longer have to retrieve information. Everything will be reported, and a first analysis will be done to set a cold balance of the situation at a given moment automatically. From there, project management will need experts who are able to investigate and develop the figures provided, from another angle. We might be needed more data analyst with Project management understanding rather than fully trained Project Managers. This will also be a game changer when evaluating risks, since the impact will be directly estimated. This concern is unexpectedly raised by the survey when seeing that 36% of the consultants we interviewed– which are professionals in Change Management and Project Management are seeing AI as threatening their future professional life. A fewer amount of job is expected to be offered to gather data, analyse them and send them to senior role for a higher critical analysis. (add team management idea)
However, this evolution is not only about increasing the standards of the Project management practitioners recruited but also their typology. Surely we can expect a higher demand for expert in project management but we also have to take into consideration the increase of the demand of “hybrid” practitioners with knowledge in machine learning and systems. According to our survey, Data Analysis is the area that will be the most impacted by AI according to 41 % of our respondents, followed by reporting activities (23%). Indeed, an algorithm will always follow the common proverb “garbage in, garbage out”. Even if it is highly sophisticated it is still “just”, for at least a medium range of time, a “narrow artificial intelligence”. It is meant to perform a simple task programmed by a human employee. What is implied with this argument is the idea is that the data will not be directly retrieved and analysed by a human officer but through a software which will be configured to perform the task as thought by a human developer. There, the precision of the analysis will be related to two aspects: the way the software will be programmed to treat data (the intelligence and the logic put into it to treat the data) and the variable inputted. From here on, it appears really interesting for a project management officer to understand and develop its own tools to adapt them to the context of project. This is the complementarity of knowledge raised by Adrian Müller “the project manager defines his or her information needs, and the data mining algorithm computes the variables.”
5.1.2 The limits of a change in the typology of PM roles
Nevertheless, the change of typology of the future project manager needs to be balanced. On the one hand, as one of the consultants we have interviewed answered related to data mining, “There are many opportunities for companies with the right blend of skills to monetise hi-volume data”. But on the other hand, we should not take that affirmation for granted and extrapolate it to every sectors and projects. Small scaled project for example might not need the implementation of a data mining or machine learning related tool. This can either happen because the amount of data available is not enough or because the data collected are too hard to analyse and the interactions involved are too complex. (ex : marketing car neuro science illégale sur le sol français.)
The promise of a more ethical project management
With Artificial intelligence is introduced the idea of a fair analysis of the data retrieved and above all, more ethical project management behaviors. Two of those consequences, to name a few, are transparent information and unbiased allocation of resources. But to which extent is machine learning expected to disrupt this topic?
Xx focus on expert skills : real need to raise knowledge. Fairness in allocation of resources: recruitment only based on experience not age or gender
Fairness in the reporting and the evaluation of employees
Limits : depend on the variable we wants to input and how weight them, + ethical behavior such as transparent practices are prompt by an external forces : the public opinion. After everything a human is choosing what to do.
Hosley, W. N. (1987). The application of artificial intelligence software to project management. Project Management Journal, 18(3), 73–75.
Couch, C. (2015). All in for AI: as funding for artificial intelligence projects surges, organizations find themselves grappling with a talent shortage and concerns about job displacement. PM Network, 29(10), 12–13.
Méndez, Y. B. & Yerkes, L. (2008). Creating lasting change: tools and techniques for effective consulting. Paper presented at PMI® Global Congress 2008—Asia Pacific, Sydney, New South Wales, Australia. Newtown Square, PA: Project Management Institute.
Vargas, R. V. (2015). Applying neural networks and analogous estimating to determine the project budget. Paper presented at PMI® Global Congress 2015—North America, Orlando, FL. Newtown Square, PA: Project Management Institute.
Parsi, N. (2017). Get Smarter: Artificial Intelligence (AI) R&D Projects Deliver Advantages. PM Network, 31(12), 16–17.
Swanson, S. A. (2017). Real Possibilities: Artificial Intelligence Has The Potential to Allow Project Managers to Focus on Higher-Value Work. The Question Is: Are They Ready? PM Network, 31(1), 62–67.
Morad, A. A. & Vorster, M. C. (1993). Network-based versus AI-based techniques in project planning. Project Management Journal, 24(1), 23–30, 36.
Kartam, N. & Levitt, R. E. (1991). An artificial intelligence approach to project planning under uncertainty. Project Management Journal, 22(2), 7–11.
John Chelliah, (2017) “Will artificial intelligence usurp white collar jobs?”, Human Resource Management International Digest, Vol. 25 Issue: 3, pp.1-3,
Everett, C. (Ed.). (2015, October 09). Artificial intelligence: HR’s role in the robot takeover. Retrieved from https://www.personneltoday.com/hr/artificial-intelligence-hrs-role-robot-takeover/Benedikt Frey, Carl ; A. Osborne, Michael. (2013). The Future of Employment: How Susceptible Are Jobs to Computerisation?. Oxford Martin. 114. 10.1016/j.techfore.2016.08.019.
A Basic Introduction To Neural Networks. (n.d.). Retrieved from http://pages.cs.wisc.edu/~bolo/shipyard/neural/local.html
Larson, E. J. (n.d.). The Limits of Modern AI: A Story. Retrieved from https://thebestschools.org/magazine/limits-of-modern-ai/