Aipedia: Structural and civil engineers
AI can help sort and interpret borehole logs and test results, but it cannot see ground between boreholes, and an engineer must set and explain the limits of any interpretation. Ground conditions drive foundation design, cost and programme, so errors here travel far.
Part of: AI in structural and civil engineering
In short
Ground investigation produces logs, lab results and in-situ tests at discrete points. AI tools can digitise logs, correlate strata and propose ground models, but they interpolate and extrapolate in ways that may look precise without being supported by the data. The Statement of Ethical Principles asks engineers to explain uncertainties and the limits of evidence. When ground turns out different from what was reported, claims often follow from contractors and clients. PI usually responds to a negligent interpretation claim, subject to the wording, and the engineer's records of how the model was built and checked are central.
Uses include extracting data from scanned historic logs, building three-dimensional ground models from borehole and trial pit data, estimating design parameters from correlations, flagging anomalies in test results, and drafting the factual and interpretive sections of ground investigation reports. On larger schemes AI can combine site data with published geological information to suggest where further investigation would add most value.
That can be useful, but the AI does not know the site history, the drilling quality or why a particular log looks odd. A geotechnical engineer does, or should find out.
Data handling matters too. Ground investigation data often belongs to the client or to a previous owner, and uploading it to a public AI tool may breach confidentiality terms. IStructE advises client permission before storing data and access controls, and Principle 1 of the Statement of Ethical Principles asks engineers to respect confidentiality and intellectual property.
The classic failure is extrapolation: a ground model that draws smooth strata between widely spaced boreholes and misses a buried channel, soft lens or old fill. AI correlations can also apply parameter relationships developed for one soil type to another. Digitising tools can misread historic logs, and a model built from misread data inherits the error. Because outputs are often visual and detailed, they can suggest more certainty than the investigation supports.
Illustrative scenario: an AI ground model shows consistent clay across a site, and pile lengths are set from it. During piling, the contractor meets a zone of soft alluvium between two boreholes. Piles have to be redesigned and lengthened, the programme slips, and the contractor claims unforeseen ground costs from the employer, who then looks to the engineer.
Unforeseen ground disputes often involve several parties, and the engineer may face a claim from the client or a contribution claim from others. PI usually responds to an allegation that the interpretation or advice was negligent. The defence typically turns on whether the investigation and interpretation were reasonable, whether further investigation was recommended, and whether the report made its uncertainties clear. Our guide to PI for civil engineers and engineer negligence claims examples explain the wider picture. Cover depends on the policy wording.
Related: AI flood and drainage modelling and AI-drafted engineering reports. Back to AI in structural and civil engineering.
Apex arranges PI for civil and geotechnical engineers. A broker reviews your cover and how your policy treats AI-assisted ground interpretation.
It can propose a model, but it cannot know what is there. Any interpolation needs engineering judgement about site history and geology, and the report should explain the uncertainty, as Principle 4 of the Statement of Ethical Principles expects.
It can save time, but check a sample against the originals, because misread depths or descriptions flow into every later step. IStructE also advises checking ownership of historic drawings, and the same care applies to historic ground data.
Yes, it is sensible. IStructE says clients should be told what AI is used for, and readers relying on an interpretive report need to know which parts were modelled by software and that a competent engineer reviewed them. It also supports the report's stated limitations.
It depends on the contract and on whether the ground was reasonably foreseeable from the information available. An engineer whose interpretation fell below a reasonable standard may face a claim, and PI usually responds to that allegation, subject to the wording.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.