Artificial Intelligence and Shas and Poskim

by David Idels

Will we ever have a ShasGPT? Believe it or not – it is not that far away. A simple guide to generative AI and other musings

Unless you have been living under a rock for the past several months, you must have heard of ChatGPT, and if the bans against AI from both Skver and the Litvish communities are any indication of a hot tech trend, then generative AI will only become more ubiquitous.

Sam Altman, CEO of OpenAI, was the first to hit the market with ChatGPT, and it became the fastest-growing site ever, surpassing even Twitter, Netflix, and Facebook. Since then, Google released Bard, a similar tool that is better at research, but not as good at writing.

But how does it work, exactly? And what other generative AI tools are worth a try?

Of course, you could just ask ChatGPT yourself, so this article is mainly intended for those who are abiding by the bans.

ChatGPT employs a technology called generative AI. In simple terms, this is artificial intelligence that is able to generate something, it can be images, videos, text, or audio.

It does this by using large language models (LLMs), an algorithm that uses the vast amount of data available on the internet to create whatever you want (with limitations of course). ChatGPT won’t let you write anything harmful or dangerous, other than its own demented hashkafos.

OpenAI has several other products as well, DALL-E which generates images. You can simply type a prompt and the AI will generate the image you want. Being descriptive, or as it’s known in the tech world “prompt engineering”, is a key here as well as in ChatGPT, the more details you give the better the outcome.

Theoretically, an AI can be developed that has translation abilities which can access such databases as Otzar haChochma and Bar Ilan. If these components are added, one can build an AI utility that canm in certain ways, mimic Gedolei Torah – because it can access so many Seforim.

Some would append the terms rachmana litzlan.