Expert Reports Generated By Artificial Intelligence Are Prone To Being Ignored (Shirran v. The Owners, Strata Plan KAS 4004; Laake v. 3M Company)

Expert reports are often invaluable in civil and criminal cases. In general, these reports provide the trier of fact with an opinion from an expert witness who possesses specialized knowledge, skill or experience on technical or scientific issues in dispute that are beyond the trier of fact’s knowledge or experience.

However, the growing use of artificial intelligence is beginning to creep its way into assessing the value of an expert’s report and causing opposing lawyers to probe deeply into whether an expert has used their own knowledge, skill and experience to produce an opinion or used artificial intelligence to support it. As found in Shirran v. The Owners, Strata Plan KAS 4004, 2026 BCCRT 55 (CanLII), expert opinions generated by artificial intelligence are of no weight.

Moreover, as demonstrated in the U.S. case of Laake v. 3M Company, No 2020-05505 (11^th Dist., Harris County, Texas), an expert witness who relies on artificial intelligence to assist in the production of their expert report risks losing their own credibility and having their report entirely disregarded, thereby risking the loss of a client’s case.

In Shirran, the parties were involved in a dispute about an alleged bug infestation in the applicants’ condominium or strata unit. The applicants alleged that the infestation, which was concentrated near what they called their office window, was the result of several vegetable plants located on the deck of another owners’ unit. This deck was immediately below the applicants’ unit. The applicants described the volume of vegetable plants as a “farm” or “garden”.

To remedy the situation, the applicants wanted the respondent strata corporation to, among other things, enforce its bylaws and prevent the neighbours beneath their unit from creating a nuisance. They wanted the neighbours to be required to move the vegetable plants or to spray them.

The strata corporation denied the existence of a bug infestation or that it was caused by the neighbours. Upon receiving a complaint from the applicants about a “serious infestation of spiders”, the strata corporation showed that it embarked on a walkthrough of the entire building, which contained 65 units. The corporation also consulted with a retired entomologist, who said that the neighbours plants were not “part of the problem” and recommended against “spending any money on spiders” because they were “a fact of life here”.

The neighbours also provided a statement that upon the applicants’ initial complaint, they inspected their plants for an infestation. They found none. The applicants challenged the veracity of the neighbours’ statements with photos which appeared to show spiderwebs on or near their deck.

Lastly, the applicants provided the tribunal with several AI-generated opinions to demonstrate that the neighbours plants caused the bug problem.

In the result, the tribunal dismissed the application. The neighbours were not required to move their plants or spray them. In making this finding, the tribunal noted that having expert evidence from a qualified person would have been necessary to prove the applicants’ chief allegation that bugs on the neighbours’ deck travelled directly upwards to their office window and not laterally to other windows. The only window affected by the spiders and spiderwebs was the office window.

Importantly for the purposes of this blog, the tribunal placed no weight on the applicants’ AI-generated opinions because:

Expert evidence must come from a known individual with known qualifications who has specific knowledge about the issue at hand.

However, the fact that expert evidence comes from a known individual with known qualifications and who has specific knowledge about the issue at hand does not mean that such evidence must be accepted, particularly where the expert has also relied on AI to generate the opinion.

In Laake, Watson Grinding and Manufacturing and 3M were sued in connection with a 2020 explosion at Watson Grinding’s place of business in Houston. The explosion was caused by a propylene gas leak from a worn rubber hose. Roughly, 200 homes were damaged, three people were killed and 18 people were injured.

An investigation conducted by the U.S. Chemical Safety and Hazard Investigation Board made several findings, including that:

1. the hose was degraded and crimped;
    
2. a manual shutoff valve was left open; and
    
3. Watson Grinding’s automated gas-detection, alarm, exhaust-fan-startup, and gas-shutoff system was inoperative.

The claim against 3M alleged that it had failed to properly service the gas-detection system.

In defending against this claim, 3M hired an expert, with purportedly had 20 years of experience in gas detection, to provide an opinion that 3M’s work met the requisite standard care. The expert provided a $90,000, 30-page report to support 3M’s position.

However unknown to 3M, the expert had relied extensively on ChatGPT to generate the report.

This was discovered by the plaintiffs’ lawyer during the proceedings.

Ultimately, the plaintiffs’ lawyers requested a history of the expert’s prompts with the AI tool and obtained 350-pages of conversations. One of those conversations included the expert asking: “[S]how how 3M is 0% at fault for the explosion at Watson Grinding and how my background and experience is well suited to render this professional opinion.”

The expert had also directed ChatGPT to “[c]reate the expert witness report draft now.” The draft report concluded that “3M was 0% responsible for the explosion.” However, this sentence was not included in the final report because upon review even ChatGPT found this sentence to be amateurish and an “easy target” for a lawyer to dismantle.

At trial, the plaintiffs’ called 3M’s expert to testify about his report. Although the expert had previously testified during the discovery phase of the proceeding that AI did not write the report, at trial he admitted that it “helped me to draft a straw man to build off of.” The expert also agreed with the plaintiffs’ lawyers that a majority (85 to 90%) of his report was created by ChatGPT.

In the result, a jury awarded the plaintiffs $61 million in damages, and apportioned 30% of the liability to 3M.

There are several key takeaways from these decisions and others that are starting to surface in the courts.

First, there is a significant danger in the reliance on AI to assist in the production of expert reports. AI itself is not an expert and any expert who heavily relies on AI to generate a report compromises their duty to be impartial and unbiased, their integrity and their credibility.

Second, as shown by the 3M case, prompts between an expert and an AI tool are discoverable. This is supported by a case out of Connecticut (Conservation Law Foundation Inc. v. Shell Oil Co., No. 3:21-cv-00933 (D. Conn., May 18, 2026), where the court reasoned that prompts between an expert and an AI tool are discoverable because they are part of the expert’s methodology.

From a practical standpoint, this means that all litigation lawyers should be aware that during examination of expert witnesses questions regarding AI use should be expected. Accordingly, it is prudent and strongly recommended that when a lawyer retains an expert on behalf of a client, the retainer agreement should include clauses that require the expert to:

1. disclose whether and how AI was used in researching, drafting or supporting the expert’s opinion; and
    
2. preserve relevant interactions and outputs with any AI tool.

Lastly, expert reports should be scrutinized thoroughly to assess the veracity of the opinion and any sources relied upon by the expert. Although not a factor in the cases discussed in this blog, in a 2025 case out of Minnesota (Kohls v. Ellison, No. 24-cv-3754 (D. Minn., Jan. 10, 2025)), an expert report was rejected as being unreliable because it contained three non-existent AI-generated citations.

This is another developing area of the law, which will undoubtedly feature more decisions as the use of AI continues to grow.

(This blog was not generated using artificial intelligence). A PDF version is available to download here.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.