The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
No Problem with Expert's Using ChatGPT to Confirm His Work
"Lehnert used ChatGPT after he had written his report to confirm his findings, which were based on his decades of experience joining dissimilar materials."
From yesterday's decision by Judge Gary Brown (E.D.N.Y.) in Ferlito v. Harbor Freight Tools USA, Inc.:
Plaintiff purchased a splitting maul (an axe specially designed for splitting wood) from defendant in 2017. Several months later, while plaintiff was hanging the maul to store it, the head of the tool detached and struck plaintiff, causing injuries to his nose and left eye. Plaintiff initiated this lawsuit in 2020, alleging that the head detached due to a design defect; defendant asserts the product failed due to plaintiff's misuse, which it contends is evidenced by a large crack in the handle….
To support his defective design claim, plaintiff seeks to offer expert testimony by Mark Lehnert, who identifies himself as a "consultant with products and liability history, extensive knowledge and experience in manufacturing and assembly, [and] mechanical and electrical engineering management." Lehnert holds no engineering degrees, yet reports extensive experience designing and manufacturing power tools, holds over a dozen patents, and has worked in management positions in engineering departments at several corporations over a period of decades.
Lehnert contends the maul used by plaintiff was defectively designed because the handle and head were weakly bound with adhesive, leading to the accident. He opines that good design requires securely attaching the head and handle by "drilling a small diameter hole through the side of the maul, into and through the handle" and placing an aluminum pin "through the head" to reduce the possibility of separation. Lehnert's report references several other mauls currently available for purchase that incorporate such a pin….
Defendant moves to preclude Lehnert's testimony, arguing that he is unqualified as an expert because he lacks engineering degrees, and his experience is limited to designing power tools rather than manual tools. Defendant further argues that Lehnert's opinion is unreliable because (i) he did not rely on any scientific, technical, or trade articles in preparing his report, and (ii) after completing the report, he entered a query into ChatGPT about the best way to secure a hammer head to a handle, which produced a response consistent with his expert opinion….
No problem, says the court:
Federal courts have grappled with the appropriateness of an expert's use of artificial intelligence to form opinions, and the validity of AI as a research tool in litigation more broadly. See Kohls v. Ellison (D. Minn. 2025) (excluding expert testimony when the expert's affidavit contained ChatGPT-generated references to non-existent academic articles); Mata v. Avianca, Inc. (S.D.N.Y. 2023) (sanctioning lawyers and law firm pursuant to Rule 11 for using ChatGPT to find non-existent cases, which the attorneys cited in a filing); Park v. Kim (2d Cir. 2024) (referring attorney to the Court's Grievance Panel for relying on ChatGPT to write a brief containing non-existent cases).
In Kohls, the expert's "citation to fake, AI-generated sources in his declaration … shatter[ed] his credibility with th[e] Court" such that his testimony would not be reliable as required by [the federal rules related to admissibility of expert evidence]. However, the Court emphasized that experts can use "AI for research purposes" given its "potential to revolutionize legal practice for the better." [Admissibility] issues arise only "when attorneys and experts abdicate their independent judgment and critical thinking skills in favor of ready-made AI-generated answers."
Here, there is little risk that Lehnert's use of ChatGPT impaired his judgment regarding proper methods for securing the maul's head to its handle. The record from the hearing reflects that Lehnert used ChatGPT after he had written his report to confirm his findings, which were based on his decades of experience joining dissimilar materials. During the hearing, Lehnert professed to being "quite amazed" that the "ChatGPT search confirmed what [he] had already opined." …
There is no indication that Lehnert used ChatGPT to generate a report with false authority or that his use of AI would render his testimony less reliable. Accordingly, the Court finds no issue with Lehnert's use of ChatGPT in this instance….
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
" He opines that good design requires securely attaching the head and handle by "drilling a small diameter hole through the side of the maul, into and through the handle" and placing an aluminum pin "through the head" to reduce the possibility of separation."
I mean, the legal point might be valid, but, geeze, what awful engineering advice. A steel pin would be cheaper and about ten times stronger, and might be reasonable as a backup retention feature, rather than a primary one.
Assuming a fiberglass rather than wood handle, adhesive is actually pretty standard. That's how my favorite splitting maul is built. But, Harbor Freight, yeah, they probably did use an inferior adhesive.
If it was a wooden handle, it should have been attached via the insertion of wedges into the end of the wood, which would split it and hence wedge it tight.
Not an expert, but have been using axes and mauls for 50 years.
And what about contributory negligence? He didn't notice that the head was almost detached from the handle?
Made me wonder how "hanging the maul to store it" could end up with "the head of the tool detached and struck plaintiff, causing injuries to his nose and left eye". I've had shovels separate from the handle, and there was plenty of warning beforehand from the looseness. It's hard to imagine using a maul such that something cracked or broke during that last swing and he didn't notice until he swung it up against a wall to hang it.
I've had an axe blade end up lose after long use, but I agree, I can't imagine one being lose enough to fall off, and not having noticed it.
Unless maybe the maul was being put away by a different person than had been using it, and they hadn't told him the head was about to fall off.
That would be two dumb people. I can't imagine picking up and handling any maul that fragile and not noticing a little wiggle. I can't imagine noticing a maul about to fall apart and telling someone else to put it away instead of setting it aside to repair or take back for a warranty claim.
Never forget that half the population is under 100 IQ.
People constantly type 'loose' when they mean 'lose' but I hardly ever see the reverse happen. But no at least I can confirm that both drive me crazy.
Neuropathy; I can't really feel if I've hit the keys hard enough to register, (Have a hard time staying at "home", too.) and if I typo by not hitting one hard enough, the computer's not going to tell me.
I fully expect to be reduced to hunt and peck within a few years.
When I pay $25 for a maul, I expect it to be the same quality as the ones used by Caltech high energy physicists. And NASA.
Aluminum is better for a device that is subject to impact forces. Steel has a tendency to work harden and can become brittle. Yes I am a Mechanical Engineer.
Two words: "galvanic corrosion."
I don't know why, but for some electrical reason, steel eats aluminum when damp, which it will be if you are splitting green wood.
It's why you have to use aluminum nails for aluminum siding, although I am not sure why you have to use aluminum nails for vinal siding. It's also why you have a chunk of zinc attached to your (wooden) boat because it will get dissolved first.
Sacrificial anodes are used with metal hulls, not wooden.
The shear strength of a steel dowel would be four times higher at the same size, and assuming the pin were going through a fiberglass handle, it's never going to be subject to forces that would harm it, the fiberglass would give up before that happened.
I suppose you could make a case for a substantially larger aluminum pin, but he specified a "small" pin! For a splitting maul?
In any event, a pin should only be a secondary retention feature, using epoxy as the primary retention is standard in axes and mauls that have fiberglass handles. I don't think I've ever seen one that did have a pin, and I've been using one since the 70's.
I don't often say this but Dr Ed is right on all counts. An aluminum pin would be astonishingly stupid for this purpose. If you must use a pin, it should be of the same material (or galvanically equivalent) as the steel head.
That said, any pin is just stupid for this purpose. If the handle is fiberglass, adhesive is standard. If the handle is wood (and it almost certainly is given the evidence that it was cracked), wedges in the end are standard. Either creates a solid non-wiggling joint between the head and the handle. A pin might keep the head from flying off for the first swing or two but if that's all you're depending on, the pin itself is going to degrade and weaken the handle with use.
This "expert" shouldn't necessarily be disqualified for using ChatGPT but should be disqualified for not being an actual expert in the topic.
Furthermore, anyone not inspecting their tools before putting them away deserves what they get.
He shouldn't even have had to inspect it. If it was loose enough to fall apart merely from hanging it on a wall, he would have had to be incredibly unaware to not notice it, and what was he doing, standing directly underneath it to hang it from rafters when it fell apart? It would be interesting to know more details.
To the extent he is otherwise an expert, I agree the fact that he queried ChatGPT does not disqualify him. But I sure hope whatever fact finder is presented with that report gives exactly zero weight to whatever text ChatGPT produced.
"could be built better" =/= "defective."
Harbor Freight = JUNK!!!
I got burned, badly, with their trailer tires -- which they sold as being for road use and then had you sign some waiver of liability on them.
Harbor Freight -- if it can be defective, it will be.
Buying any product that requires a liability waiver is contributory negligence.
(yes, including alleged vaccines)
I’m not aware of a vaccine liability waiver. There is a vaccine court and a statutory framework that provides limited liabilities for vaccine injuries. But that was created by Congress for the same basic reasons we have worker’s compensation programs that limit liability for on-the-job injuries.
The government signed the vaccine liability waiver for you by putting it into law.
" . . . arguing that he is unqualified as an expert because he lacks engineering degrees . . . "
Rensselaer Polytechnic Institute, founded in Troy, NY in 1824, awarded the first degree in Civil Engineering in 1835, although at the time it was only a one-year technical finishing program.
So before 1935 there were no qualified engineers?
Don't tell the Romans.
Or Greeks.
Or Egyptians.
2025 is going to be memorialized and the year the legal profession self destructed.
I am a Harbor Freight loyalist and frequent customer, but I have realistic expectations about their tools. At their prices, anything beyond a couple of uses is a bonus/good fortune.
ETA: For anyone who cares, we're talking about a $25 maul here, and that's before the famous & frequent HF discount coupons.
I buy Harbor Freight when I only expect to use a tool once a year or one time. Any extended use is a bonus. There's just as much a place for Harbor Freight junk as expensive stuff which will last years of heavy duty use. It's pretty elitist for all those folks to applaud making cheap tools unavailable to people who can only afford cheap tools.
I have a splitting maul, but it's a decent one, because I have a fireplace, and end up splitting quite a bit of wood.
Mostly sweetgum, which you have to whack about a dozen times more times to split than you do oak... I actually had to buy a set of splitting wedges for the stubborn pieces.
Just in time for weekend chainsaw work: HF coupon for 200lb capacity portable work stand, $19.99.
https://go.harborfreight.com/coupons/2025/04/183537-38778/
FWIW -- (Mainly) just for fun, I recently asked Gemini to draft a trust agreement with somewhat nonstandard terms, trustee powers, etc. It did a very impressive job, incl. language citing relevant state statutory provisions re: delegation and other stuff I needed. With further queries and final human editing, took about 5 min start to finish.
When all the hype around Chat GPT and LLM in general started, I decided to run some tests. I picked a legal topic I have taught and asked it questions I have previously used on mid terms and finals. I also fed it UBE essay questions on that topic. Surprisingly, the results were quite good. Good enough to pass a bar exam, or land in the 'bottom half but passing' section of a law school class.
I have since asked it to generate motions on topics I deal with frequently. Not for my own use, but to see what such a motion from opposing counsel might look like. On a fairly common, fairly straightforward topic, it generated a result that could be used in a motion with only light editing. Obviously it couldn't generate a 'statement of facts' that would work for a particular case. But it certainly generated something that could be used as 'argument' that correctly stated the standard of review and the applicable case law (with cites) from the correct jurisdiction. If people are routinely submitting AI written motions with a hallucinated cites and case law (and from following EV it appears they do), the people using the LLMs must be quite sloppy indeed. Or they must be litigating outside of their area of expertise. I can't imagine asking an LLM to author something, seeing it generate a case I've never heard of, and just going "Huh, well look at that. I've never heard of that case even though I'm supposedly an expert in this area. But if the LLM says it's true it must be." That's not just bad lawyering. That's bad personing.
I can't get past the part where they recommend an aluminum pin to secure an axe head. That makes me think they don't know what they're doing. Not only is it a third material but it isn't a common method for securing axe heads nor is aluminum a material that consistently has high sheer or shock load resistance. If you really feel you need a pin steel would have been a more obvious choice.
Never.ind that the idiot hit himself in the face with an axe head because he couldn't be bothered to know what condition it was in before apparently trying to handle it high enough that failure caused injury. This type of personal negligence lawsuit should just be dismissed based on the plaintiff being a complacent idiot.
I put the question to chatgpt just now and got this answer:
When selecting a metal for a pin to attach a splitting maul handle to the axe head, you're looking for something strong, durable, and resistant to wear and impact. The two primary options are:
1. Carbon Steel
Strength: Carbon steel is a common and reliable material for making pins due to its strength and hardness. It’s tougher than some alloys and can hold up to the intense impact of a splitting maul.
Durability: It’s relatively wear-resistant, which is important for a pin that will be exposed to high stress. With a good heat treatment, it will last.
Workability: Carbon steel is relatively easy to work with when forging or machining.
Recommendation: Look for medium to high carbon steel (around 1045 to 1060) for a good balance of strength and toughness.
2. Stainless Steel
Strength: While stainless steel is highly resistant to rust and corrosion, its strength can vary depending on the alloy. Some stainless steels (like 17-4 or 416) are strong enough for this application.
Durability: Stainless steel resists corrosion, which can be important if the maul is used in wet or humid conditions.
Workability: Stainless can be more difficult to work with and may require more effort to machine or forge, but it’s a good option if you want added protection against rust.
Recommendation: 304 or 316 stainless would be sufficient, but ensure that the pin is hardened enough to resist deformation.
3. Tool Steel (e.g., A2 or D2)
Strength: Tool steels like A2 and D2 are designed for high wear resistance and can handle the stresses of heavy impact. These steels are tough and have excellent edge retention, making them good candidates for pins.
Durability: These steels are generally very durable, although they can be more difficult to forge and machine than standard carbon steel.
Recommendation: Tool steels offer excellent durability but might be overkill unless you're aiming for a high-end, long-lasting result.
Conclusion:
For general use (and ease of maintenance), 1045 carbon steel is a great option. It's strong, tough, and relatively easy to work with.
If corrosion resistance is important (e.g., for outdoor or marine environments), consider 304 stainless steel.
For ultimate durability and toughness, tool steel (A2 or D2) would be ideal but requires more effort to machine.
As you can see, aluminum isn't even listed.
I've been an expert witness in product liability cases and medmal.
I'd hate to think that I was getting that money to run an AI.
I know less than zero about product liability, so just wondering: Since the head fell off when he tried to hang it on wall for storage, rather than when actually using it to split wood, is that relevant? I assume it's an arguable point on both sides, but I demand a definitive answer. 🙂
Interesting extended discussion on the relative merits of differently manufactured splitting mauls...among people who, I assume, must the parents that the young adults in the the series of insurance commercials are turning into.
I resemble that remark.
P.S. Cargo shorts are cool ... and practical.
I started reading the facts before looking at the parties and first thought was it Harbor Freight. Made of Chinesium.