APM PMO Interest Network Conference - James Martin-Young - Intelligent Project Prediction
0 likes617 views
The document discusses the adoption of AI in project management, focusing on Intelligent Project Prediction (IPP) and its potential to enhance project outcomes through machine learning and data analysis. It outlines the capabilities and limitations of AI, as well as the future of AI-enabled project delivery by improving decision-making, resource management, and risk mitigation. A case study demonstrates significant improvements in project predictions and financial outcomes after implementing IPP.
APM PMO Interest Network Conference - James Martin-Young - Intelligent Project Prediction
- 1. The Big PMO Adoption Thursday 11 July 2024 etc.venues Manchester apm.org.uk/event
- 2. Intelligent Project Prediction and how it can enable improved outcomes with Artificial Intelligence and Machine Learning James Martin-Young Head of Digital & Data, MiGSO-PCUBED
- 3. Intelligent Project Prediction and how it can enable improved outcomes with Artificial Intelligence and Machine Learning James M-Y
- 4. AI – Our/My Journey 2019 2023 2024 2026 We’re on a journey, learning how AI in PPM can enable improved outcomes for projects, programmes & portfolios
- 5. AI – The Headlines From the news headlines it is easy to believe AI is something new and threatening AI isn’t new, but Generative AI is exploding in utility AI is generating fake news AI-Generated Art Wins Prestigious Art Competition AI-Powered Deepfakes Raise Concerns About Identity Theft AI passes MBA, bar and medical examinations AI-Powered Recruitment app stopped because of bias AI Job Displacement Raises Fears of Mass Unemployment AI's Rapid Advancement Raises Concerns About Society's Ability to Adapt
- 6. AI – Focus Areas Creates new content (text, images, music). Used in chatbots, virtual assistants, and content generation. AI that learns from data without explicit programming. Used in fraud detection, spam filtering, and medical diagnosis. Enables computer-human language interaction. Applied in machine translation, text summarization, and question answering. Generative AI NLP (Natural Language Processing) Machine Learning (ML)
- 7. AI – Being Responsible It is important to recognise some of the current limitations It’s Not Truly Intelligent Think of it like a super version of predictive text. It works on 1 word at a time and predicts the most likely word based on what you asked, what it knows about you, what is has already learned and what it has already said. Produces Different Answers It can give different people different answers and it tends to be positive (it is a people pleaser) Sustainability AI uses significant amounts of electricity. Gartner says we should be responsible in understanding computing powers impact on sustainability targets Requires Training It can only tell you about what it has been trained on. All Data Is Equal It thinks all data is equal, it does not understand aspects such as privacy or property rights Not Truthful It does not give you the truth it gives you what is most probably the truth from what it has learnt and its’ calculations. We should be conscious that any insights that drive PPM decision making need to be ratified by a human
- 8. Future AI-enabled Delivery • The future delivery ecosystem will be shaped by the advent of technological advances. • Some of the technology’s already here, and by combining a series of Artificial Intelligence (AI) applications, we can leverage this technology to drive delivery success. • Today we’ll talk about Intelligent Project Prediction and where we are working to improve project outcomes. Virtual agent Selecting Work Machine-learning optimisation to select the best-balanced portfolio in line with strategic ambition 2 Planning & Scheduling Predictive analytics to forecast delivery timelines, budgets and resource allocation 3 1 Delivery Management LLM-enabled chatbots able to synthesise queries and provide relevant reporting 4 Stakeholders & Suppliers Generative production and delivery of comms, with data- driven insight on stakeholder sentiment and recommended interventions 5 Dynamic Scheduling Event-triggered routines dynamically re-scheduling activities to optimise resources and outcomes 6 Suggested Interventions LLM-generated recommendations on actionable and effective interventions based on historic performance 7 Managing Risk Scenario planning driven by predictive analytics to forecast the impact of various risk- mitigating interventions 8 Learning Lessons Ingesting and mining documentation using NLP with sentiment analyses, generatively producing machine and human readable insight summaries 9 Sponsoring Investment Prediction-driven analysis preventing the funding of projects destined for overrun without conditions being set prior to approval
- 9. Piloting AI-enabled Delivery Virtual agent Selecting Work Machine-learning optimisation to select the best-balanced portfolio in line with strategic ambition 2 Planning & Scheduling Predictive analytics to forecast delivery timelines, budgets and resource allocation 3 Managing Risk Scenario planning driven by predictive analytics to forecast the impact of various risk- mitigating interventions 8 Sponsoring Investment Prediction-driven analysis preventing the funding of projects destined for overrun without conditions being set prior to approval 1 Learning Lessons Ingesting and mining documentation using NLP with sentiment analyses, generatively producing machine and human readable insight summaries 9
- 10. M|P Manchester – 3rd April 2023 AI-Enabled Delivery Insights
- 11. Why Intelligent Project Prediction (IPP)? 11 Late running or overspending projects Over stretched resources Recurring delivery issues Siloed project management Challenges Course correction and improved project delivery Efficient workflows and resource management Transform persistent challenges Minimum project management standards Traditional Solutions Project forecasting informed by robust prediction Improved resource management workflows Prioritised intervention and improved project selection Integrated systems with high data maturity Augmented Solutions
- 12. What is Intelligent Project Prediction A project to prove… Using Machine Learning Trained on real historic Financial, Schedule & Other PPM Data To prove robust predictions Enable interventions to produce project improvement
- 13. What is Intelligent Project Prediction (IPP)? Predict Project Outcomes 96% accurate predictions on project financial & schedule data Significantly Reduced Project Overrun & Overspend Intervene before projects go off course Improved Decision Making for Executives ML models continue to learn and provide insights Significant Portfolio ROI on IPP 18% savings opportunity in live case study Project Portfolio Investment Decisions Predict from Business Case to hone investment decisions Reduced Impact From Human Forecasting Remove optimism and politics from your project portfolio AI Driven accurate predictions with rationale at portfolio level with live project outcomes
- 14. IPP: How does it work? 14 Current &Historic Project Data • Existing PPM Data Top-Down Data • Data Model & Quality Assessment Common Data Model Implementation & Quick Wins • AI &ML Learning algorithms trained on existing data (3 months) Analytics Engine Dashboard, Insights & Intervention Identification • User friendly intuitive visualisations Improve &Transform • Recommended Improvement Strategies and digital assets • Portfolio Transformation • Enhanced Decision Making
- 15. IPP Case Study – Global Logistics Challenge Greater certainty over delivery outcomes to increase project success and provide Exec knowledge where to focus their limited time. Portfolio of c300 Projects, significant %failing to deliver to originally planned time and cost. (Infra, Business Change, Technology) Solution Provision of an up to date, dynamic Intelligent Project Prediction tool that predicts individual project outcomes and compares to what is being reported, and identifies core drivers for project success / failure
- 16. IPP Case Study Outcomes • In the first year after the deployment of IPP, there was: A reduction of the projects which overspent their budget from 33% to under 6% (an 83% reduction) A reduction in the level of budget overspend from those projects which did overspend down from 74% of the budget to 48% of the budget. A prediction accuracy of 95%+ Before IPP After IPP Subsequent Phase Whilst deploying to wider user base, integrating IPP with lessons learned and within the project approvals/sales stage - driving efficiency savings throughout the life- cycle.
- 17. What’s Next? Confirm ROI & benefits in real world environments Add new capability such as Chatbot & Lessons Expand portfolio and work with new customers Integrate with Gen AI& Big Data for further intelligence
- 18. Today’s presenter 18 James Martin-Young Head of Digital UK MiGSO-PCUBED