The Kavod of AI

We are trying to navigate a new world of artificial intelligence, especially its impact on individuals and society. But while the technical strides (or even leaps) forward in the world of AI continue to advance, the concerns it raises are deeply human: Will people's jobs, or even their whole career field become irrelevant? With easy access to information and generative AI, will people still value deep thought and hard work? As large language models (LLMs), such as Claude and ChatGPT, and AI-generated videos become more prevalent, will we be able to distinguish reality from an AI-generated fiction?

Over the last few months, I've been teaching and reflecting on how AI is being deployed. I've also heard strong feelings about how it's being used and whether it should be avoided. There's a generational and professional difference in how AI is perceived: those in the STEM fields are finding AI to be an incredible tool for things like medical diagnoses and coding, while those in the humanities and creative fields are pushing hard against using it for writing or teaching purposes. The question is not as simple as whether AI is good or bad, or even whether AI should be used. Instead, we need to think about how and when it's appropriate to utilize AI, especially if we want to make sure we keep our humanity.

Rather than providing specific guidelines and policies that become irrelevant and outdated within months (if not weeks), perhaps it may be better to provide a simple heuristic we can use to begin reflecting on its use. Just as Hillel taught us to "strive to be human" (Pirkei Avot 2:5) to guide our behavior, we also can draw on his summary of the Torah: "What is hateful to you, do not do to another. That is the whole Torah. Now go and study" (Babylonian Talmud Shabbat 31a). This is a powerful framework because it allows us to have a "gut-check" on how we act. It won't provide particular answers, but it can provide a baseline to study the nuances, rules, policies, and specific recommendations of AI use.

For me, the most helpful heuristic is kavod, respect. When it comes to a specific question or use case, it's worth asking, "Is this respectful?" While this plays out in a variety of ways, there are three realms where the framing of kavod has helped me navigate the question of using AI and LLMs in writing and teaching.

Kavod for the work of those who came before us 

A statement from the Babylonian Talmud, Megillah 15a, has become quite relevant since the adoption of LLMs: "Whoever reports a saying in the name of the one who said it brings redemption to the world." Giving credit to those who have created written works, music, art, and more is a core value in Jewish tradition. Speaking b'shem omro, in the name of the original speaker, gives respect to the time and effort they spent developing the idea and bringing it into the world. LLMs like ChatGPT and Claude generally don't cite specific sources, often hallucinate texts, and are built on previous works without the original creators' consent or compensating them.

At the same time, creators don't want their ideas to live only in their minds. While there is still major debate about how creators should be honored, LLMs can find new and unexpected sources, discover inspiration, or make connections that may not be obvious. Asking an LLM to provide the exact source (then checking and citing the source!) may actually uncover the writings and ideas of otherwise unknown contributors.

Kavod for the audience

When we hear a sermon or read a d'var torah, we expect to hear something thoughtful and original. If we discover that it was generated by an LLM, we feel disrespected, like our time was wasted or the piece was simply dashed off. Why would we want to listen to or read something that is generic and interchangeable?

However, AI can also be tremendously individualized and personalized, so it can be helpful for sharpening ideas, tightening language, or clarifying complicated ideas. On one level, writing is crucial for developing thinking. But sermons, curricula, and divrei torah aren't ultimately meant for the creator, they are meant for the audience. So, if an LLM can help a writer strengthen or deepen their ideas for the audience, we could argue that its use gives more respect to those who will read or listen to it.

Kavod for time

Like any new technology, it's debatable whether AI saves time or simply creates more work. But if it can be used for many emails or administrative tasks that are formulaic and repetitive, AI can allow people to devote their time to more productive and meaningful pursuits.

Yet efficiency is not the ultimate goal of life. It's not just about saving time, but how that saved time is used. Shabbat, for example, is designed to get us to interact with other people, which can sometimes be complicated and force us to deal with interpersonal friction. Sometimes, we don't want to simply get through a task but deepen a relationship; and that requires both quality and quantity time.

It will continue to be increasingly easier to rely on AI for much of our work and our lives. But we also need to ensure that we keep respecting our humanity, ourselves, and those we interact with. There will be more complex questions, nuanced discussions, and challenges facing our use of AI. However, if we can hold onto kavod as a heuristic, we can then do as Hillel said and "go out and study."