For my second test-drive of Google Bard, I queried: “what is the test for single asset real estate in a bankruptcy case?” I know something (quite a lot—I am about to submit an article for publication on this topic) about this topic. I ran this question twice, on different days.
Bard responded differently each time. I ran a redline to compare the answers (it is attached).
I am disconcerted by the variation in the two answers on the two days. For instance, the first answer says there is an 80% subtest for “substantially all the gross income” element of the definition of SARE, and the next answer says it is 70%! Which is it, 80% or 70%, Bard?
More significantly, I do not find such a numerical test regnant in the case law.
For another example, Bard’s 2d answer says SARE cannot be "a collection of multiple properties. For example, a shopping center would be considered a single property, but a group of apartment buildings would not.” While I dislike the SARE rules in the Bankruptcy Code [spoiler alert: my new article makes the case for treating such debtors more equitably], I believe that multiple properties can constitute a single “project.” Furthermore, neither a shopping center nor a group of apartment houses is necessarily SARE—the key is whether the debtor is engaged in entrepreneurial activities on and with the real property, as my article will show.
Also unsettlingly, I see today that Google Bard offers a choice of answers. At the top right of the screen showing the answer is an option to “View other drafts.” The “other drafts” are varying—I would say divergent—answers to the same question.
So I decline to test Google Bard further. I conclude that lawyers, legal historians, and laymen should not rely on its answers to either historical or legal questions.
