I thank Chaim Saiman, David Zvi Kalman, and L.M. Sacasas for their thoughtful responses to my essay.
I infer from each of the responses that at least one aspect of my essay bears clarification. I posited that certain central features of the Jewish tradition might serve to protect Jewish communities and individuals from some of the potential dangers of AI, but that this would require that Jewish tradition adapt successfully to rapidly changing circumstances. Some of the concerns raised by all three respondents could be addressed through a brief elaboration on my views regarding the adaptiveness of Jewish tradition—specifically, halakhah, Jewish law.
In Darwinian evolution, mutations are random and those relatively few mutations that improve the ability to reproduce in a given environment will eventually proliferate. In cultural evolution, the role of random mutations is played by deliberate innovations, some of which help a culture to flourish and some of which fail. It’s hard to know in advance which are which because societies are delicate organisms and it’s not easy to anticipate which innovations will lead to unintended harmful consequences.
In the specific case of halakhah, innovation in response to changing circumstances happens through subtle interaction between rabbinic experts and the wisdom of crowds. It is always concrete—addressing specific cases (in the spirit of common law, as Chaim Saiman notes)—and incremental—the specific cases addressed are borderline issues. It is also subject to rabbinic disputes and communal variation, so that views and practices compete and ultimately the best-adapted win. In halakhah it also isn’t always clear in advance what the winners will be, so that a certain amount of humility is called for. The very slowly expanding body of entrenched views and practices, along with discussions of their rationale vis-à-vis rejected alternatives, forms the basis for collective intuitions regarding new borderline cases, and so on.
While the analogy between Darwinian evolution and halakhic development is illustrative, it’s important to note the unique deliberative and communal aspects of halakhic evolution, distinguishing it from the purely natural and random processes observed in biology. This deliberative development of halakhah is neither random nor perfectly planned, but rather somewhere on the continuum between these poles.
With that in mind, let us consider the critiques of my essay. I noted that precisely because halakhah is an evolving system, it is less likely to fall prey to AI-inspired fads. I further claimed that Shabbat laws would adapt in a manner that would preserve the “tuning-out” quality of the Sabbath that will be even more important in the age of immersive AI. But Saiman notes correctly that I didn’t only say that certain Shabbat laws would ultimately adapt in some hard-to-predict way. In fact, I assumed that our intuitions regarding these matters are so clear that this process could be subtly engineered to ensure efficient adaptation of Shabbat laws to anticipated technological developments. But this assumption, even if true, implies that these particular laws might be outliers in my own evolution analogy. At the very least, as Saiman correctly observes, it is unlikely that we will have similar clarity regarding novel matters in other areas of Jewish law, such as kashrut. Point well taken.
L.M. Sacasas, writing as a Christian, notes, among many other observations, that all religious people share certain concerns about the effects of AI on religion. These include the fear that one salient benefit of AI for religion—AI’s ability to pre-digest difficult religious texts—could lead to shallow engagement with those texts. This fear is indeed justified, as I wrote, and I’m pleased to hear that a thinker outside the Jewish tradition sees eye to eye with me on this point.
This point, too, can be better explicated in light of my comments about adaptiveness. One important difference between, on the one hand, engaging with texts in which differing views on legal matters are considered and weighed and, on the other hand, merely consulting an AI to obtain an oracular apodictic ruling is this: only through engagement with the arguments for and against different views and practices that have been considered in the past can we develop the intuitions required for attempting to resolve novel borderline cases. For the process of adaptation to be more efficient than random mutation and natural selection, we need to cultivate the necessary intuitions through serious study.
Finally, David Zvi Kalman is concerned that my suggestions regarding how Jews might contribute to the broader conversation regarding the perils of AI are too parochial. Kalman would like to raise the level of discourse among Jews on this topic in numerous ways, including transcending narrow halakhic considerations. As he puts it: “the top priority for Jewish leaders must be formulating basic moral attitudes from which particular rules can eventually be derived” so that they might “stake a moral claim in a global conversation.”
I’m sympathetic to Kalman’s ambition to serve as a light unto nations and appreciate that if we are to do so, we must speak in words and concepts that are intelligible and meaningful outside our own community. I would only caution that if we are to do so authentically as Jews, we must offer insights that emerge organically from that which is distinctive to Jewish tradition. The basic moral attitudes from which particular rules can be derived must themselves emerge from the bedrock of established rules. We must be cognizant of the fact that if all we offer are generic bromides teased selectively out of ambiguous folklore, we are merely speaking as Westerners or as people of a particular political persuasion. It is not Kalman’s intention, of course, that we do anything of the sort, but I fear that untethering our thinking from the halakhic framework risks such a result. That result might have some value, as far as it goes, but if we are to enrich the dialogue without the appearance of chauvinism, we must take care not to stake claims, moral or otherwise, to borrowed goods.
On two key points, I am glad to see that I am in agreement with all three of my respondents: first that we would be well advised to direct our attention elsewhere than the fantastical nightmare scenarios of AI taking over the world and enslaving humanity, and second that we cannot afford to take lightly the disruptions this new technology will bring. I am convinced that Judaism has the resources to confront these disruptions, if we Jews are prudent about how we use them.