João Freitas

The following is a retrospective on the features of LLM (Large Language Models) in assisting both developers and customers as an interface (a blackbox that allows consumption). It goes on explaining how LLM can be integrated as a translator in programming, creating new compression algorithms for specific data domains, and how it can also be used to simulate interactions with other interfaces, such as a computer virtual machine.

https://matt-rickard.ghost.io/ai-interfaces/


The current generation of LLMs uses natural language as an input/output. This is convenient (and impressive) for human interaction, but what about computer-to-computer communication?

Emulating wire formats

With a simple prompt, GPT-3 can easily be coerced into accepting and outputting JSON. You can even use the prompt to specify the schema for the API responses you want (you can simply give it a few example responses). This makes it easy to build traditional systems around model inference.

Maybe this solves ETL and the M:N API problem, to some degree. Fuzzy mappers can handle small unexpected changes in an API response. Of course, maybe this introduces more opportunities for hidden problems.

Encoders/decoders

In addition, GPT-3 is quite good at learning (or creating) encoders/decoders. This means it could plausibly generate good compression algorithms that match the data. For general-purpose models, this might not always give the right results – it would look something more like a probabilistic data structure like a bloom filter. But fine-tuned GPT-3 encoders and decoders might go a long way into being efficient ways to exchange data.

Emulating APIs

ChatGPT has been successful in hallucinating APIs,

The problem, of course, is hallucination – i.e., there’s no guarantee that the results are referentially correct. But it does post the question – if you have a fuzzy interface or a simple enough interface, you might be able to replace the backend with something like GPT-3.

#reads #matt rickard #ai #llm #chatgpt