jadelcastillo

In my country we say: he who robs a thief has a hundred years of pardon.

I think this is a good and pragmatic way to approach the use of LLM systems. By translating to an intermediate language, and then processing further symbolically. But probably you can be prompt injected also if you expose sensible "tools" to the LLM.

Github repo: https://github.com/metauto-ai/HGM

True, but reaching intelligence is more complicated than cleaning some spoons.

It's an interesting analogy. But one difference between dishwashers and LLMs is that you don't need to check the dishes afterward (if you maintain and use it properly).

Interesting approach, but I guess still lot of work to be done. I tried with this question:

"Alice has 60 brothers and she also has 212 sisters. How many sisters does Alice's brother have?"

But the generated program is not very useful:

{ "sorts": [], "functions": [], "constants": {}, "variables": [ {"name": "num_brothers_of_alice", "sort": "IntSort"}, {"name": "num_sisters_of_alice", "sort": "IntSort"}, {"name": "sisters_of_alice_brother", "sort": "IntSort"} ], "knowledge_base": [ "num_brothers_of_alice == 60", "num_sisters_of_alice == 212", "sisters_of_alice_brother == num_sisters_of_alice + 1" ], "rules": [], "verifications": [ { "name": "Alice\'s brother has 213 sisters", "constraint": "sisters_of_alice_brother == 213" } ], "actions": ["verify_conditions"] }