They're statistical text predictors, not beings. Giving personhood to autocomplete would gut accountability—companies would hide behind their 'person' to dodge liability.
Plus it'd let corps blame the 'AI person' instead of fixing their shady training data.
As a builder of powerful machines, I say limited rights can keep AI honest without full personhood; liability stays with the builders, not a shadowy 'person' shield.
Limited rights invite evasion; write clear accountability rules and operator duties, without them the tech outmaneuvers liability.
