3M, Nighthawk engineering, and the ChatGPT expert report

A consultant retained by 3M via Nighthawk engineering (paid "$90,000") asked ChatGPT to 'show how 3M is 0% at fault for the explosion at Watson grinding' and used the output as an expert opinion in litigation about a gas explosion. The AI produced a '30 page report', self‑rated it (said it was 97/100), and included the explicit line 'from a technical and standard of care standpoint, 3M is 0% responsible for the January 24, 2020 explosion.' The consultant submitted that report to the court without adequate independent analysis.

Opposing counsel noticed a suspicious five‑page 'citation overlay' and asked under oath whether ChatGPT was used. The judge allowed the defense to download and read 350 pages of the consultant's ChatGPT history into the record. The transcript records that the chat history included an uploaded photograph of the gas detector (the central object of the case) and unrelated prompts such as t‑shirt designs. Those discoveries undermined the consultant's credibility.

The jury awarded "$61 million" and apportioned "30% of the responsibility" to 3M, not the '0%' the AI‑generated report had declared. The speaker frames this as a cautionary tale about overreliance on generative AI: professionals who outsource judgment to ChatGPT risk producing reports that are legally discoverable, ethically compromised, and professionally disastrous.

Practical takeaways urged by the speaker: do not let ChatGPT draft definitive expert conclusions; retain independent technical judgment; avoid uploading critical evidence or confidential photos to third‑party models; and rebuild core skills rather than delegating them. He also reiterates a business rule he uses: 'The code is like 10% of a software business. The rest is marketing, distribution, branding, storytelling, social media, and just sheer raw perseverance.'