​
Overview

How can we increase a model’s performance on open-ended tasks? Role prompting, or persona prompting, assigns a role to the model. Roles can be:
  • specific to the query: You are a talented writer. Write me a poem.
  • general/social: You are a helpful AI assistant. Write me a poem.

​
More Role Prompting

To read about a systematic approach to choosing roles, check out RoleLLM. For more examples of social roles, check out this evaluation of social roles in system prompts. To read about using more than one role, check out Multi-Persona Self-Collaboration.

​
Example

<?php
require 'examples/boot.php';

use Cognesy\Instructor\Extras\Sequence\Sequence;
use Cognesy\Instructor\StructuredOutput;
use Cognesy\Utils\Arrays;

class Company {
    public string $name;
    public string $country;
    public string $industry;
    public string $websiteUrl;
}

class GenerateLeads {
    public function __invoke(array $criteria, array $roles) : array {
        $criteriaStr = Arrays::toBullets($criteria);
        $rolesStr = Arrays::toBullets($roles);
        return (new StructuredOutput)->with(
            messages: [
                ['role' => 'user', 'content' => "Your roles:\n{$rolesStr}\n\n"],
                ['role' => 'user', 'content' => "List companies meeting criteria:\n{$criteriaStr}\n\n"],
            ],
            responseModel: Sequence::of(Company::class),
        )->get()->toArray();
    }
}

$companies = (new GenerateLeads)(
    criteria: [
        "insurtech",
        "located in US, Canada or Europe",
        "mentioned on ProductHunt",
    ],
    roles: [
        "insurtech expert",
        "active participant in VC ecosystem",
    ]
);

dump($companies);
?>

​
References

  1. RoleLLM: Benchmarking, Eliciting, and Enhancing Role-Playing Abilities of Large Language Models
  2. Is “A Helpful Assistant” the Best Role for Large Language Models? A Systematic Evaluation of Social Roles in System Prompts
  3. Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration