Expression of interest: AI roundtable

Stream 1: Case studies

Explore how AI is being used in structural engineering, both in design and non-design roles:

Non-design applications

• Reducing admin overhead: What specific time-consuming tasks (emails, reports, scheduling, document control) are being automated using AI?
• Coding and automation: Many engineers are using AI to write scripts and automate workflows, how widespread is this, and what are the most common applications?
• Data processing and knowledge management: Can AI help engineers retrieve past project knowledge more efficiently?
• AI for regulatory compliance: Is anyone using AI tools to track and align with evolving design codes and regulations?
• Assisting with quality control: How are firms using AI to check calculations and identify potential errors in documentation?

 Design applications

• Reality vs. marketing hype: Many companies claim to have 'AI-powered' tools, how do we differentiate real advancements from exaggerated claims?
• Are engineers designing with AI? Do we have real examples of structural engineers using AI for generative design, topology optimisation, or automated detailing?
• AI for early-stage design: Is AI helping with feasibility studies, site selection, or conceptual massing?
• Integration with BIM and parametric design: How is AI enhancing workflows within platforms like Revit, Rhino/Grasshopper, and Tekla?
• Bridging design and construction: Are AI tools being used to optimise constructability, sequencing, or modularisation?

Stream 2: Ethical practice and legal considerations

Address the challenges of fairness, accountability, and compliance in AI adoption:

Ethical considerations

• Impact on hiring and skills development: Are firms hiring fewer junior engineers because AI can handle basic tasks?
• Graduate training and deskilling risk: Should new engineers rely on AI, or do they need to learn fundamental skills without it first?
• Bias and transparency: AI models can reinforce existing biases, how do we ensure structural decisions remain objective?
• Accountability for errors: If an AI-generated design fails, who takes responsibility, the engineer, the software developer, or neither?
• The human role in AI workflows: Should AI only be used as a tool to assist decision-making, rather than replace engineers’ judgment?

 Legal considerations

• Data ownership and intellectual property: If AI is trained on past projects, who owns that knowledge, the software provider, the firm, or the original designers?
• Privacy concerns: Even using AI for emails or design discussions might expose sensitive data, how should firms manage this risk?
• Siloing of sensitive information: What are the risks of using AI tools for projects under NDAs (e.g., nuclear, defense, confidential commercial work)?
• Compliance with industry standards: Are current design codes and liability frameworks equipped to handle AI-assisted work?
• Regulatory oversight: Should there be legal requirements for auditing AI-generated designs, and if so, who enforces them?

Stream 3: Safety concerns

Explore risks related to reliability, accountability, and oversight of AI-driven processes:

Ensuring trust in AI-generated outputs

• The black-box problem: How do we ensure that AI-generated designs are transparent and verifiable?
• Hallucinations and overconfidence: AI can confidently produce incorrect results, how do we safeguard against this?
• Verification processes: Should firms implement mandatory human checks for AI-assisted calculations, and how rigorous should they be?
• AI in risk-critical structures: Would you trust an AI-designed bridge, nuclear plant, or skyscraper?

Liability and accountability

• Who takes responsibility for errors? If an AI-assisted design fails, who is legally responsible, the engineer, the software developer, or neither?
• Standards for AI in engineering: Should there be industry-wide guidelines for auditing and validating AI-assisted designs?
• The role of human oversight: How much human intervention is necessary before an AI-generated design can be considered safe?
• Automation vs. professional judgment: Should AI recommendations ever override an experienced engineer’s decision, and if so, in what scenarios?