By Michael Lustig
FaithInvest’s exploration of how ChatGPT can be applied towards drafting faith-consistent investing guidelines is extremely interesting and illuminating. (To see FaithInvest's original post, click HERE.) Although I’ll admit to a bias against using this new tool as a shortcut, it truly does seem to have a lot of utility; at the same time, it falls-flat in a few respects.
Wearing my Adjunct Professor ‘hat’ I teach an Impact Investing course, and the recent rise of ChatGPT coincided with a Final Exam that I was giving. To my own astonishment (though probably no surprise to those who have been closely following media reports) the ‘bot scored an 88 (out of 100) in a side-by-side test, placing it solidly in the ‘A-/B+’ range. This both disturbed and unnerved me, and yes, I’ll be (highly) reformulating the types of questions that I ask of my students when I next give this course!
That said, although the exam answers were more ‘composed’ than what a student would probably produce, they were still primarily, in essence, correct. In a highly similar manner, the Jewish values identified by ChatGPT are very credible, and could certainly serve as the nucleus of reasonable investment guidelines. This isn’t terribly surprising, given how ChatGPT works, since it’s very likely that the system ‘Pre-trained’ (the ‘P’ in ‘GPT’) from existing Jewish investing frameworks like JLens and others, as it scoured the internet in its “learning” process.
What ChatGPT lacks is human judgment and creativity. The output that was produced is essentially a list, with no context, no source-referencing, and no examples. It’s highly-caveated so as to not take a strong position even while it’s (presumably) advocating for a certain course of action. There’s no prioritization, no action-items, and no acknowledgment of where there is overlap/duplication in the principles that it has identified. In short, there’s no soul.
As FaithInvest noted in its post, a more robust query might yield a deeper result, though there are still inherent limitations to what a ChatBot can do. Using the tool as a first-cut is reasonable, as the output reflects a very deep-dive into the rich resources that may be found on the internet, but it should be done with the understanding that this is indeed just a starting-point; scrutinizing/editing/augmenting the ChatGPT result with a human-touch is both desired & necessary.
About the Author
Michael Lustig is Adjunct Professor of Finance at NYU | Stern School of Business, and a Lecturer at several other academic institutions, teaching full-time elective courses for MBAs in Fixed Income, Capital Markets, and Impact Finance. He retired after a 25-year career on Wall Street, mainly at the asset management firm BlackRock, where he was a Managing Director.
Michael is an advocate for the adoption of Impact Investing in the Jewish community, active through both Jewish Federations of North America and the Jewish Funders Network, and is author of A Guide to Jewish Impact Investing. He is an active investor in ImpactTech ventures, and sits on the National Advisory Board for Israel to the G7's Global Steering Group on Impact Investing. Michael lives in Manhattan with his wife Rachel, and their three wonderful children.