AI Application for Geoprofessionals

Ramtin Tajeddin, Negin Yousefpour, Wai Leung Ng, and Philip Tsang

March 13, 2026

AGS VIC chapter is pleased to announce the Panel Discussion on AI application for geoprofessionals.

The potential application and experience of using AI will be discussed by the panel. The event aims to spark conversation, identify the risks and opportunities of this ever-growing technology.

Questions

What are the differences between academic advancement in ML and AI, and implementation of them in industry by practitioners? What are the roadblocks for implementation?
What is the trend of changes in the number and competencies required in geotechnical engineering due to AI?
How necessary is big data in AU and NZ in implementation of AI and ML?
What should be the course to get it started.
In your current role, where does AI actually touch your day-to-day geotechnical work today?
What was the first task you trusted AI with in geotechnical practice, and what made you comfortable doing so?
Do you see AI primarily as a productivity tool, a decision-support tool, or something closer to a junior engineer—and why?
How are you using AI with subsurface data—logs, lab results, CPTs, monitoring data—and what works well versus poorly?
Have you integrated AI into any repeatable workflows or templates (e.g. report drafting, risk registers, scope generation), can you give us some example?
Can AI meaningfully assist with engineering judgement, or is it still limited to supporting tasks around it?
Where do you personally draw the line between AI assistance and engineering responsibility?
How do you validate or verify AI-generated outputs before relying on them professionally?
Do you think current professional standards, QA systems, and insurances are adequate for AI-assisted geotechnical work?
If an AI-assisted decision contributes to a failure, where does accountability realistically sit?
What skills will define a strong geotechnical engineer in 5–10 years in an AI-enabled industry?
Does AI raise the floor, lower the ceiling, or both, for junior geotechnical engineers?
What is one AI capability you expect to materially change geotechnical practice in the next 3–5 years?
What advice would you give to a geotechnical engineer who is sceptical of AI but feels pressure to adopt it?
Are we training the next generation of geotechnical engineers—or quietly deskilling them?
What is the role of the industry bodies, AGS EA etc in preparing the application of AI
If AI consistently outperforms average engineers in certain tasks, is it unethical not to use it?
Is the biggest AI risk technical failure—or misaligned commercial incentives?

Correction

Ramtin Tajeddin, correction: to question regarding the liability of autonomous vehicle (AV) companies.

“My previous remarks suggested that these entities consistently evade responsibility by shifting blame to human operators. While this was a prevalent concern during the experimental and “supervised” phases of 2023, the regulatory landscape has evolved significantly as of 2026.

As we have moved into full Level 4 commercial deployment, the following changes have rendered my previous claim outdated:

Statutory Responsibility: In major markets like California and the UK (under the Automated Vehicles Act 2024), the law now identifies the manufacturing or operating company as the legal “driver” or “operator” when the autonomous system is engaged. This removes the legal burden of the driving task from the human occupant.
Strict Accountability: New frameworks, such as the UK’s “Authorised Self-Driving Entity” (ASDE) model, make the corporate entity responsible for traffic infractions and collisions caused by the software, ensuring that passengers are not held liable for the vehicle’s technical decisions.
Insurance and Enforcement: Regulators now mandate high-level liability coverage (often $5M+) for driverless fleets, and as of 2026, law enforcement in several jurisdictions can now issue “notices of noncompliance” directly to manufacturers for traffic violations.

When a human is not driving, a collision is often treated as a “product failure” rather than a “driving error.” This brings the case under the umbrella of product liability law. Victims can argue that the vehicle’s software or hardware was “defective” in its design or manufacturing. For example, if a LiDAR sensor fails to detect a pedestrian in a crosswalk due to a known software bug, the manufacturer can be held strictly liable, meaning the plaintiff does not necessarily have to prove that the company was “careless,” only that the product was dangerous and failed during normal use.

I believe it is important for the published record to reflect that while companies still utilise traditional legal defences (such as comparative negligence regarding third parties), they can no longer legally deflect the primary responsibility for the “driving task” to the human passenger in a fully autonomous vehicle.”