Can AI Help Explain Bankruptcy Law?
A newbie parses Alix v. McKinsey with the help of an LLM1
I am a machine-learning engineer who maintains a database of bankruptcy case law, at the behest of practicing attorneys. This database powers ChatLaw, an AI service that helps bankruptcy attorneys get up to speed on less familiar legal matters. Currently in beta, ChatLaw fetches answers to bankruptcy-related questions from relevant case law. Sometimes I get curious about the new cases we add to the database. Enter Jay Alix v. McKinsey & Co. I took a peek at the new 60-page opinion, filed on 8/18 at the SDNY, and was immediately assailed by a Pandora's box of bafflements.
Firstly, how does this case relate to bankruptcy? Neither of the two parties was a debtor or creditor. So I asked ChatLaw2, at https://www.chatlaw.us:
I dictated: What is the significance of McKenzie [sic] versus Jay Alex [sic] in Bankruptcy law3
ChatLaw's response: The case of Jay Alix v. McKinsey & Co (Case: 18-CV-4141, Date: 8/18/2023) in the Southern District of New York is significant in bankruptcy law for several reasons4.
Firstly, it highlights the importance of full disclosure in bankruptcy proceedings. The case alleges that McKinsey & Co., a bankruptcy consulting firm, concealed connections to parties with an interest in the debtor estates, thereby obtaining bankruptcy consulting assignments under false pretenses. This is in violation of the requirement for bankruptcy professionals to file declarations disclosing any connections to the debtor estate and confirming that they are disinterested persons.
Secondly, the case underscores the concept of proximate causation in Racketeer Influenced and Corrupt Organizations (RICO) Act claims. The plaintiff, Jay Alix, founder of a competing bankruptcy consulting firm, AlixPartners LLP, alleges that McKinsey's fraudulent practices caused his firm to lose considerable revenue. The court initially dismissed Alix's claims due to a failure to adequately allege proximate causation. However, the Second Circuit later vacated and remanded this decision, indicating that the issue of proximate causation in such cases is complex and may require further factual investigation.
Thirdly, the case demonstrates the application of the doctrine of collateral estoppel, or issue preclusion, in bankruptcy proceedings. McKinsey argued that Alix's claims related to the Alpha Natural Resources bankruptcy were barred by collateral estoppel because they had been previously litigated and rejected. However, the court rejected this argument, noting that the initial challenges were dismissed on standing grounds, not on the merits.
In conclusion, the case of Jay Alix v. McKinsey & Co is significant in bankruptcy law as it underscores the importance of full disclosure, explores the concept of proximate causation in RICO claims, and illustrates the application of the doctrine of collateral estoppel.
Chat Log 1
My dictated question had some spelling errors; but ChatLaw understood my intentions. Its response clued me in that the case is about a disclosure and retention issue amongst competing professional services firms in bankruptcy proceedings. But it also brought new questions. I had no idea what collateral estoppel meant. Can ChatLaw help again?
I typed: What is the doctrine of collateral estoppel?
ChatLaw: The doctrine of collateral estoppel, also known as issue preclusion, bars relitigation of any factual or legal issue that was actually decided in previous litigation between the parties, whether on the same or a different claim. This doctrine is recognized and can be applied in bankruptcy court proceedings (Grogan v. Garner, 498 U.S. 279, 284, 1991).
Under federal law, the application of collateral estoppel requires that:
1) the identical issue was raised in a previous proceeding; 2) the issue was actually litigated and decided in the previous proceeding; 3) the party had a full and fair opportunity to litigate the issue; and 4) the resolution of the issue was necessary to support a valid and final judgment on the merits (Ball v. A.O. Smith Corp., 451 F.3d 66, 69, 2d Cir. 2006).
In New York, the doctrine of collateral estoppel bars a party from relitigating an issue which has previously been decided against him in a proceeding in which he had a fair opportunity to fully litigate the point (Khandhar v. Elfenbein, 943 F.2d 244, 247, 2d Cir. 1991).
In California, collateral estoppel applies if:
1) the issue sought to be precluded from relitigation must be identical to that decided in a former proceeding; 2) this issue must have been actually litigated in the former proceeding; 3) it must have been necessarily decided in the former proceeding; 4) the decision in the former proceeding must be final and on the merits; 5) the party against whom preclusion is sought must be the same as, or in privity with, the party to the former proceeding (Harmon v. Kobrin (In re Cantrell), 329 F.3d 1119, 1123, 9th Cir. 2003)5.
Relevant cases include: Parklane Hosiery Co. v. Shore, 439 U.S. 322, 326 (1979), Scarbrough v. Purser (In re Scarbrough), 836 F.3d 447, 456 (5th Cir. 2016), Kremer, 456 U.S. at 487 (quoting Montana v. United States, 440 U.S. 147, 164 n.11 (1979)), Lindsey v. Prive Corp., 161 F.3d 886, 891 (5th Cir. 1998) (quoting Montana v. United States, 440 U.S. at 164 n.11), Schwager v. Fallas (In re Schwager), 121 F.3d 177, 181 (5th Cir. 1997), Park v. Chang (In re Park), 271 F. Appx 398, 400 n.5 (5th Case 17-03446 Document 30 Filed in TXSB on 11/19/18 Page 36 of 100 37 Cir. 2008) (quoting Garner v. Lehrer (In re Garner), 56 F.3d 677, 679 (5th Cir. 1995)).
Chat Log 2
I find the above explanation of collateral estoppel quite clear. In fact, the concept seems so basic that I suspect all attorneys know it. Nevertheless, ChatLaw's response includes historical detail and jurisdictional nuances that might be useful to an attorney whose case touches upon it.
Getting back to Alix v. McKinsey, I was curious whether the opinion I uploaded supported one side's argument more than the other. The concluding order read: McKinsey's motion to dismiss is GRANTED in part and DENIED in part. I couldn't tell how meaningful that was for either side. So I asked ChatLaw's opinion on the opinion:
I typed: How strong is Jay Alix's suit against McKinsey in his allegation of violation of the RICO statute?
ChatLaw: Jay Alix's suit against McKinsey & Co. alleges that McKinsey's pattern of filing false and misleading declarations constitutes racketeering activity under the Racketeer Influenced and Corrupt Organizations (RICO) Act, 18 U.S.C. 1961-1968. Alix brings three substantive RICO claims under Section 1962(c) and one claim of RICO conspiracy under Section 1962(d).
The strength of Alix's suit can be evaluated based on the court's response to McKinsey's motion to dismiss Alix's claims. The court granted the motion in part and denied it in part, indicating that some of Alix's claims were found to have merit while others were not.
The court found that Alix had sufficiently alleged a direct relationship between McKinsey's injurious conduct and Alix's injury, reasoning that McKinsey entities likely would not have received their bankruptcy assignments had they been found ineligible because of fraud. This was a key point in the Second Circuit's decision to vacate and remand the case (Alix v. McKinsey & Co., Inc. (Alix I), 404 F. Supp. 3d 827 (S.D.N.Y. 2019), vacated and remanded by 23 F.4th 196 (2d Cir. 2022) (Alix II)).
However, the court dismissed some of Alix's claims against individual defendants due to the statute of limitations and lack of specific allegations. For example, the court dismissed Alix's claims against McKinsey Holdings because it found no specific acts by McKinsey Holdings or by executives acting within the scope of their employment with, or for the benefit of, McKinsey Holdings.
The court also dismissed Alix's claims of money laundering as a predicate act because Alix did not allege that the pre-petition payments it received from the debtor were the proceeds of unlawful activity.
In conclusion, while some of Alix's claims under the RICO statute were dismissed, others were found to have merit and will proceed. The strength of Alix's suit against McKinsey thus appears to be mixed based on the court's rulings.
Chat Log 3
To me, the above reads like a decent, even-handed summary.
Having peeked at a number of other opinions, I am finding bankruptcy law more complex than machine-learning, where the math and algorithms are comparatively clear cut. ChatLaw, made possible by the advent of ChatGPT, allows a non-expert like me make sense of bankruptcy law in a way not possible a year ago. The attorneys helping me with ChatLaw say that even experienced legal professionals stand to benefit. Every attorney's circle of expertise has a border, a border that routinely gets breached by case specifics. If you'll pardon the pun, Alix v. McKinsey is a case in point. Not all bankruptcy attorneys will be familiar with the RICO statute.
In AI Chatbots Are Useless for Bankruptcy Lawyering6, the author Daniel III argues that "Lawyering is a dynamic process of human interactions that the AI devices cannot understand or perform." I tend to agree. What I believe AI can do is make legal knowledge, embedded in both case law and the statutes, substantially more accessible to humans—both non-experts and lawyers.
* * *
Addendum:
The risk of AI hallucinations is a question you likely have queued up for me. 'Hallucinations' are when the AI fabricates plausible but false facts to provide a good-looking but incorrect answer. ChatLaw uses a design that naturally avoids hallucinations. This design has it formulate an answer only from the case law in its database. And if no such answer is available, it has it say so. I worked with practicing attorneys to confirm that the system works as intended, and that the responses reproduced here, including the citations therein, are accurate.
Large-Language-Model. An example would be ChatGPT. ↩
The chat logs reproduced in this article have not been edited or altered in any way (other than identifying me as the questioner and ChatLaw as the respondent). ↩
I sometimes spell words wrong or cite incorrectly to test the robustness of the system. Users may not always enter a question in proper form. ↩
The attorneys I have worked with have stressed that the opinions I cite should be in Bluebook form whenever possible. You will see most of the citations from ChatLaw do conform to Bluebook form. When that is not the case, while we work to improve the program, any citation is typically easily confirmed through a web search or other online services practicing lawyers have access to.
Separately, the attorneys who have reviewed ChatLaw's answers say that ChatLaw does not always list the best case, but it almost always provides a relevant case. As disclaimed on ChatLaw's webpage, all citations should be independently confirmed before being used. ↩To ensure the reliability of the citations, the attorneys I work with manually confirmed the citations. They noted a mistake in this citation (incorrect caption but otherwise correct), and went back to the original case to see what happened. As it turned out, the judge in the case appears to have miscited the case, and so the citation picked up by ChatLaw was miscited accordingly. ↩
AI Chatbots Are Useless for Bankruptcy Lawyering. Josiah M. Daniel III. Vinson & Elkins L.L.P.; UT Austin, Dept. of History.August 13, 2023 ↩