AI Safety Research

Long Ouyang

Thesis Research

longouyang@post.harvard.edu

Project: Democratizing Programming: Synthesizing Valid Programs with Recursive Bayesian Inference

Amount Recommended:    $99,750

Project Summary

One goal of artificial intelligence is valid behavior: computers should perform tasks that people actually want them to do. The current model of programming hinders validity, largely because it focuses on the minutae of how to compute rather than the goal of what to compute. An alternative model offers hope for validity: program synthesis. Here, the user specifies what by giving a small description of their goal (e.g., input-output examples). The synthesizer then infers candidate programs matching that description, which the user selects from.

One shortcoming of synthesizers is that they are truthful rather than helpful: they return answers that are literally consistent with user requirements but no more (e.g., a requirement of “word that starts with the letter a” might return just “a”). By contrast, human read more deeply into requirements, divining the underlying intentions. Helpfulness of this kind has been intensely studied in the linguistic field called pragmatics. This project will investigate how recent developments into computational modeling of pragmatics can be leveraged to improve program synthesis, making it easier to write programs that do what we want with little to no special knowledge.

Technical Abstract

One goal of artificial intelligence is valid behavior: computers should perform tasks that people actually want them to do. The current model of programming hinders validity, largely because it focuses on the minutae of how to compute rather than the goal of what to compute. An alternative model offers hope for validity: program synthesis. Here, the user specifies what by giving a small description of their goal (e.g., input-output examples). The synthesizer then infers candidate programs matching that description, which the user selects from.

One shortcoming of synthesizers is that they are truthful rather than helpful:  they return answers that are literally consistent with user requirements but no more (e.g., a requirement of “word that starts with the letter A” might return just “a”). By contrast, human read more deeply into requirements, divining the underlying intentions. Recent work in computational psycholinguistics that we can capture this ability through user modeling — maintaining a model of how the user purposefully selects examples to convey information. This project will investigate how these psycholinguistic insights can be used to make synthesis more valid.