[cs702]

After a quick/superficial read, my understanding is that the authors:

(a) induce an LLM to take natural language inputs and generate statements in a probabilistic programming language that formally models concepts, objects, actions, etc. in a symbolic world model, drawing from a large body of research on symbolic AI that goes back to pre-deep-learning days; and

(b) perform inference using the generated formal statements, i.e., compute probability distributions over the space of possible world states that are consistent with and conditioned on the natural-language input to the LLM.

If this approach works at a larger scale, it represents a possible solution for grounding LLMs so they stop making stuff up -- an important unsolved problem.

The public repo is at https://github.com/gabegrand/world-models but the code necessary for replicating results has not been published yet.

The volume of interesting new research being done on LLMs continues to amaze me.

We sure live in interesting times!

---

PS. If any of the authors are around, please feel free to point out any errors in my understanding.

[skepticATX]

I have not yet read the paper, but based on this description it seems like it provides grounding in the context of the training data, which is kind of the rub with current LLMs to begin with, right? We don't have a set of high quality training data that is completely unbiased and factual.

[agnosticmantis]

> … which is kind of the rub with current LLMs to begin with, right?

No, the bigger problem with current LLMs is that even with high quality factual training data, they often generate seemingly plausible nonsense (e.g. cite nonexistent websites/papers as their sources.)

This is by design imo; they’re trained to generate ‘likely’ text, and they do that extremely well. There’s no guarantee for faithful retrieval from a corpus.

[novaRom]

Important addition to your partially right statement: "they’re trained to generate ‘likely’ text" is they are trained to produce most probable next word so that the current context look as "similar" to training data as possible. Where "similar" is not "equal".

[cs702]

I'd describe it as grounding the model with a formally specified symbolic world model.

[andsoitis]

Humans’ experience and understanding of the world around them isn’t limited to a symbolic representation.

It remains to be seen whether you can truly be an effective intelligence with understanding of the world if all you have are symbols that you have to manipulate.