Wicked Problems & AI Solutions: Navigating Construction's Next Frontier with RICS & Autodesk

The construction industry is standing at a pivotal juncture, with Artificial Intelligence promising to reshape everything from daily tasks to complex project delivery. But how do we navigate this rapidly evolving landscape intelligently? We recently had the pleasure of hosting Anil Sawhney, Head of Sustainability at RICS, and Jugal Makwana, Senior Executive for Industry Transformation and Strategic Open Ecosystem at Autodesk, on the Project Flux podcast. They dived deep into the themes of their co-authored RICS paper, "Wicked Problems in Construction, How the Intelligent Use of AI Can Help," offering profound insights into AI's potential and pitfalls.

Anil kicked off by perfectly capturing the current zeitgeist: imagine walking into an office five years ago with a "black box" promising to distill wisdom from all your project data. You'd have been laughed out. "But that’s where we are at," he stated. This new reality, however, brings its own set of anxieties.

What Keeps AI Experts Up at Night?

For Anil, the concerns are both personal and professional. He candidly shared his reliance on tools like Grammarly (recently blocked by his organization, much to his chagrin!) and the nervousness that comes with it. "Would we slow down in development of critical thinking, problem-solving skills if there is so much heavy reliance on chat GPT...?" he pondered, especially regarding students who are now explicitly asked to use AI in their assignments. This highlights a crucial tension: leveraging AI for efficiency versus fostering fundamental human skills.

Jugal echoed this sentiment, noting that the current AI trend is "forcing people to think how would we transform," moving beyond mere transition. The hype around AI is compelling everyone, from architects to owners, to question its impact on their work and data. "AI is starting that conversation or the hype in AI is kind of putting a pressure in everyone to think about how is it going to impact them..." Jugal observed. On a personal note, he’s found AI immensely helpful for tasks like creating a curriculum to learn Dutch and planning trips, showcasing its broad utility.

Intelligent Use of AI: Beyond the Hype

The RICS paper defines the "intelligent use of AI" as solving problems efficiently and effectively. Anil outlined three critical dimensions for evaluating any AI use case:

1. Complexity of the problem: Is AI truly needed, or can simpler methods suffice?

2. Availability of data: Is there sufficient, relevant data to train or inform the AI?

3. Additionality: Can AI genuinely perform better or offer more than a human expert? If not, it might be the wrong tool for the job.

   
   

Poor or indiscriminate AI use often involves applying it to problems that could be solved with better data analytics, or even, as James pointed out, better leadership and process discipline. Anil clarified the distinction: data analytics might involve statistical techniques like regression models, while AI often acts as the "black box." A good application? Analyzing 100 historical contracts for risk patterns or sifting through a million unstructured site inspection records – tasks where AI’s ability to handle volume and complexity offers clear additionality. "Are you gaining value by the use of AI?" Anil urged listeners to ask. "But I think evaluating complexity, data availability, and additionality may be another way of..." ensuring its appropriate application.

RAG, Data Standardisation, and the "Black Box" Challenge

Retrieval Augmented Generation (RAG) is a framework many are exploring to ground AI in specific data. However, Jugal cautioned that its effectiveness is still developing. Current RAG use cases are often document-driven, struggling to capture the deep, tacit knowledge embedded in projects. "It needs the right context to be able for it to retrieve it," Jugal emphasised, adding that success still hinges on Anil’s three factors, plus a continued "invest[ment] in standardising the data."

This sparked a key debate: the role of data standardisation. While generative AI can handle unstructured data better than previous technologies, Jugal firmly believes that "standardisation is still the key for us to succeed in this" if we want to trust the outputs of the AI "black box."

Yoshi then raised a crucial point about mechanistic interpretability, reminding us that AI is trained, not programmed, and its internal workings can be opaque. This led Anil to discuss the critical need for Explainable AI (XAI). "If it remains a black box, I think trust is going to be in deficit," he warned, especially in a contract-driven sector like construction. While generic foundation models are improving, the magic, Anil believes, still lies in combining them with robust RAG that incorporates company-specific knowledge and standard operating procedures. And above all, the principle of "Human In The Loop (HITL)" remains paramount. "Can’t do it without human in the loop," Anil stated unequivocally. "I don’t think James would be happy for his team to send out a cost estimate through CHAT GPT, which nobody has looked at."

The Rise of AI Agents and Changing Communication

The conversation naturally flowed to the rise of AI agents – tools like Manus that can autonomously perform complex tasks, from creating cost plans to deploying websites. This power brings the risk of human laziness. Jugal acknowledged this, suggesting AI will primarily take over mundane tasks like report writing, freeing humans for more engaging work. A new essential skill is emerging: "How do you manage agents?"

This will fundamentally shift how project information is communicated and consumed. Lengthy reports may give way to interactive, queryable formats. "AI is going to change the way we consume information, the way we interact with information," said Jugal. Anil concurred: "Fundamentally I feel we are going to be living in a chatbot world, even in projects." The implications are vast, even extending to how marketing might evolve, with Anil citing Perplexity’s CEO: soon, we might be "marketing to agents, not to humans."

AI for Wicked Problems vs. Leadership & Process

Anil brought the discussion back to the core of their paper: focusing AI on genuine "wicked problems" – those complex, multi-faceted challenges that no single entity can currently solve. This means consciously avoiding the application of AI to issues that simply require better leadership, collaboration, or process discipline.

The marathon effort to establish data standards, like IFC, must continue alongside our newfound ability to leverage unstructured data. "Progressing both unstructured and structured data simultaneously is going to be crucial," Anil stressed, to avoid creating new inefficiencies down the line. It’s also vital not to get distracted by AI’s “shiny objects” at the expense of quality and rigorous human oversight.

Jugal highlighted Autodesk’s journey towards becoming a platform provider, aiming to enable the "golden thread" by connecting data silos. This brings the long-standing concept of Common Data Environments (CDEs) back into sharp focus. "Organisations have now started realising and questioning, what is my data ecosystem? Where am I storing my data?" he noted. After 20+ years, the cultural shift towards shared, standardised data is gaining new urgency, driven by AI’s potential to unlock its value.

The Future of Tech Providers, Consultants, and Continuous Learning

Addressing concerns about big tech potentially displacing consultants, Jugal believes there's room for both. Autodesk, he explained, is committed to Trusted AI principles and sees its role as a platform enabler, with consultants and contractors remaining crucial for project delivery. Anil added that while tech stacks might fragment initially as companies adopt new AI tools, the 80/20 principle (80% platform-based, 20% specialised tools) will likely reassert itself, pushing platforms to become more AI-first.

Jugal’s own move from consultancy to Autodesk was driven by the "sheer potential that AI and technology have in enabling this transformation."

As for staying current, Jugal’s advice is to "spend 10 minutes learning something new every day," staying curious, humble, and adaptable. Anil views AI as uniquely transformational because, unlike previous technologies like BIM, it can permeate every aspect of an organisation. "I think AI may be like yoga for your organisation," he mused, "because it can impact your mind, body, and soul." This requires both personal enablement frameworks and overarching organisational AI strategies.

Navigating Construction’s Next Frontier

The insights from Anil Sawhney and Jugal Makwana paint a clear picture: AI offers a profound, almost limitless opportunity for the construction industry. However, its successful integration hinges on intelligent application, ethical considerations, robust data strategies, unwavering human oversight, and a commitment to continuous learning and adaptation. The journey is just beginning, but by focusing on solving true wicked problems and fostering a culture of informed curiosity, the industry can navigate this next frontier with confidence.

Dive deeper into these critical topics by downloading the full RICS paper, "Wicked Problems in Construction, How the Intelligent Use of AI Can Help," available at https://www.rics.org/news-insights/wbef/wicked-problems-in-construction-how-the-intelligent-use-of-ai-can-help .

Reflect on your organisation’s AI readiness and data strategy. And, as always, subscribe to the Project Flux newsletter at https://projectflux.beehiiv.com/  for the latest insights into the future of projects.

You can also connect with our guests on LinkedIn:

Anil Sawhney (https://www.linkedin.com/in/anilsawhney ) and

Jugal Makwana (https://nl.linkedin.com/in/jugalmakwana ).