Will AI Replace Geotechnical Engineers? The Ground Beneath Your Feet Still Needs Human Judgment
AI can analyze soil data faster than any engineer, but the judgment calls that keep buildings standing require human expertise. Here is what the numbers say.
Before any skyscraper rises, any bridge spans a river, or any tunnel burrows through a mountain, a geotechnical engineer has to answer one fundamental question: will the ground hold? It is a question that demands crawling into test pits, interpreting soil samples that behave differently from anything in the textbook, and making judgment calls that carry the weight — literally — of everything built on top of them. AI is getting remarkably good at crunching the data that informs those judgment calls. But making them? That is a different story entirely.
Our data shows that geotechnical engineers face an overall AI exposure of 40% and an automation risk of just 15/100 in 2025. [Fact] That puts them among the more AI-resistant engineering specialties, and the reason is straightforward: this is a profession where the physical world constantly surprises you, and no dataset captures the full complexity of what lies beneath the surface. The Bureau of Labor Statistics projects +4% growth through 2034, [Fact] with approximately 62,800 professionals earning a median salary of ,890. [Fact] Modest growth, strong compensation, and low automation risk — a stable combination.
Where AI Meets the Subsurface
The three core tasks of a geotechnical engineer reveal a clear hierarchy of automation potential, and the pattern is one we see across all field-intensive engineering disciplines.
Analyzing soil and subsurface data has the highest automation rate at 58%. [Fact] This is where AI makes its strongest contribution. Machine learning models can now process borehole log data, classify soil types from cone penetration test results, predict settlement behavior from historical foundation performance data, and run probabilistic analyses on slope stability faster than any human analyst. AI-driven geotechnical software can correlate data from dozens of boreholes across a site and generate preliminary subsurface models that would have taken engineers days to build manually.
But the 58% comes with an important caveat: soil is not a manufactured product. It is a natural material with infinite variability. The AI model that works perfectly for alluvial clay deposits in Houston may be wildly inaccurate for glacial till in Boston or volcanic ash in Seattle. Every geotechnical dataset is local, and every local dataset has gaps. The engineer who recognizes that a particular test result does not make sense — who questions the data rather than accepting the model output — is providing exactly the judgment that AI cannot replicate.
Designing foundation and earth retention systems sits at 32% automation. [Fact] AI-assisted design tools can optimize pile configurations, generate preliminary retaining wall designs, and run parametric analyses that compare dozens of foundation alternatives in the time it used to take to evaluate three. Generative design algorithms are beginning to explore the solution space for earth retention systems in ways that human engineers would never have time to attempt.
However, foundation design is not just an optimization problem. It involves navigating building codes that vary by jurisdiction, coordinating with structural and architectural teams whose requirements constantly evolve, and making conservative engineering choices where failure is not an option. When you are designing the foundation for a hospital in a seismic zone, the cost of getting it wrong is measured in lives, not dollars. That ethical weight and professional responsibility remains irreducibly human.
Conducting field inspections and site assessments has the lowest automation rate at just 15%. [Fact] This is the bedrock of geotechnical engineering — literally. Walking a construction site, observing excavation conditions, testing soil bearing capacity in real time, and making on-the-spot decisions about whether conditions match the design assumptions. No drone, no sensor, no AI model can replace the engineer who looks at an exposed cut face and recognizes that the soil conditions have changed from what the borings predicted.
The gap between theoretical exposure (57%) and observed exposure (23%) in 2025 [Fact] is a 34-point gap — one of the wider gaps in engineering. Geotechnical firms, particularly smaller consultancies, have been slow to adopt AI tools. The software is expensive, the learning curve is steep, and many firms rely on the accumulated experience of their senior engineers more than they rely on algorithmic optimization.
The Engineering Sector Context
Compare geotechnical engineers to their engineering neighbors. Civil engineers face somewhat higher overall exposure because more of their work involves design calculations and project management documentation that AI handles well. Construction managers see different AI exposure patterns centered on scheduling and resource optimization. But geotechnical engineers share the same fundamental advantage as all field-intensive engineering specialties: the more time you spend on-site rather than at a desk, the more AI-resistant your work is.
By 2028, we project overall exposure will reach 55% and automation risk will climb to 25/100. [Estimate] The analytical tasks will see the most change as AI-driven geotechnical software becomes more accessible. But the field inspection and foundation design tasks will evolve gradually.
What This Means for Your Career
If you work as a geotechnical engineer, the data paints a reassuring picture.
Invest in your field skills. The 15% automation rate on field inspections is your most durable career asset. Every hour you spend on construction sites — observing soil behavior under load, developing your ability to read subsurface conditions visually, building experience with different geological formations — makes you harder to replace. Senior geotechnical engineers who can walk a site and immediately identify potential problems are irreplaceable, and that skill comes only from years of field experience.
Master the AI analytical tools. The 58% automation rate on data analysis means these tools are your competitive advantage, not your enemy. Engineers who can combine AI-driven subsurface modeling with their own professional judgment will produce better designs faster. Learn the leading geotechnical AI platforms — they will make your borings, your test data, and your experience more powerful.
Specialize in complex conditions. Seismic geotechnical engineering, deep excavation support, tunneling in mixed-face conditions, foundation design in karst terrain — these are the subspecialties where the soil behaves in ways that defy simple models, and where human expertise commands premium billing rates.
Build interdisciplinary skills. Geotechnical engineers who understand structural engineering, environmental regulations, and construction methods at a practical level are more valuable than those who work in isolation. The best foundation designs come from engineers who understand how the building above will actually behave.
The ground beneath our buildings, bridges, and roads is endlessly variable, endlessly surprising, and endlessly in need of engineers who can interpret it. AI will process the data faster, but someone still needs to stand in the test pit and decide what the data means.
See the full automation analysis for Geotechnical Engineers
This analysis uses AI-assisted research based on data from the Anthropic labor market impact study (2026), BLS Occupational Outlook Handbook, and our proprietary task-level automation measurements. All statistics reflect our latest available data as of March 2026.
Related Occupations
Explore all 1,000+ occupation analyses at AI Changing Work.
Sources
- Anthropic Economic Impacts Report (2026)
- Bureau of Labor Statistics, Occupational Outlook Handbook, Civil Engineers (2024-2034 projections)
Update History
- 2026-03-29: Initial publication with 2025 actual data and 2026-2028 projections.