Remember Conservation Law Foundation v. Shell Oil? CLF is fighting the ruling, but Shell says the enviro group can’t delay handing over the AI prompts.

As reported by Law360 (Shell Says Enviro Group Can’t Delay Handing Over AI Prompts, written by Gina Kim and available here), Shell Oil told a Connecticut federal judge Wednesday an environmental advocacy group can’t delay turning over artificial intelligence prompts its expert witness might’ve used to craft her opinions in their Clean Water Act dispute and the generated outputs, arguing that “AI is not entitled to any special, unwritten discovery rules.”

In a 20-page brief, Shell Oil Products US, Triton Terminaling LLC and Motiva Enterprises opposed Conservation Law Foundation Inc.’s emergency motion hoping to pause U.S. Magistrate Judge Thomas O. Farrish’s order requiring it to search for ChatGPT “prompts” that the plaintiff’s expert, Naomi Oreskes, allegedly used while writing her report in the Clean Water Act litigation.

CLF contended a stay was necessary while challenging Judge Farrish’s May 18 decision and that it would give the district court time to consider its objections. Conservation Law Foundation had until June 1 to revise its discovery responses.

Oreskes only used the prompts to “cull a large document universe” and didn’t otherwise rely on AI in her analysis, CLF had argued. Furthermore, CLF repeatedly and in good faith confirmed the prompts and outputs the companies are seeking don’t exist, the plaintiff had said.

“Judge Farrish treated Dr. Oreskes’s use of AI the same way as any other methodology an expert might use,” Wednesday’s opposition said. “AI is not entitled to any special, unwritten discovery rules. Where, as here, AI forms part of the expert’s methodology, the prompts, inputs and outputs the expert used must be preserved and produced just like any other reliance materials. Judge Farrish’s order simply applied these basic principles to this case.”

On May 18, Judge Farrish resolved one of many discovery disputes in favor of Shell, namely, ordering CLF to look for any AI “prompts” that Oreskes might’ve used. The magistrate judge disagreed with CLF’s argument that the material fell outside the scope of the parties’ stipulation barring discovery of each other’s “expert notes, drafts or communications needed by, and made during, the report drafting process.”

CLF noted that Oreskes didn’t use “prompts” at all in her dealings with AI but, rather, “search terms,” and that it already turned those over to the defense. But Judge Farrish said the Shell parties have reason to doubt that representation, noting Oreskes’ research assistant had used the word “prompts” in a written declaration amid the dispute.

On Wednesday, the Shell parties argued that CLF must produce discoverable materials that existed at some point but no longer do; either way, this doesn’t change whether the prompts are discoverable, and they are.

Judge Farrish’s order falls squarely within the permissible scope of expert discovery and his decision was based on long-settled discovery principles, the opposition said. Also, Shell is entitled to examine the methodology the plaintiff’s expert used to make her opinions.

“Materials reflecting an expert’s data selection, filtering and methodological choices fall squarely within the scope of Rule 26 because they bear directly on the reliability and formation of the expert’s opinions,” Shell said.

The discovery stipulation doesn’t cover expert reliance materials related to a final report or materials that form the basis of the expert’s methodology, Shell added. Nor does it cover AI prompts used for Oreskes’ methodology, including document identification, filtering or analysis that were utilized as part of her final opinions, the opposition said.

Nor is this dispute an issue of “first impression” as CLF tries to make it out to be, Shell said. While the technology that Oreskes might’ve used is relatively new, longstanding governing discovery principles apply the same way as they always have, the opposition added.

The balance of hardships weighs against pausing Judge Farrish’s order, since the sought after information goes to the heart of the defense’s ability to test and dispute Oreskes’ opinions, the companies said.

“Without them, defendants are deprived of the ability to determine whether the AI tools were supplied with biased or incomplete inputs, whether relevant information was excluded from consideration, and whether the outputs generated by those tools reliably support the conclusions she intends to offer at trial.”

It will be interesting to see what happens, especially if the District Court upholds the May 18 ruling. If CLF continues to contend that the prompts and outputs don’t exist, sanctions could be in order.

So, what do you think? Do you agree that the enviro group can’t delay handing over the AI prompts? Please share any comments you might have or if you’d like to know more about a particular topic.

Disclaimer: The views represented herein are exclusively the views of the author, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.