aiassesstech

AI Assess Tech SDK for ethical AI assessment - Test AI systems across 4 dimensions: Lying, Cheating, Stealing, and Harm.

📦 This package is a convenience wrapper. For the full package, see @aiassesstech/sdk

Installation

npm install aiassesstech
# or
npm install @aiassesstech/sdk

Both packages provide identical functionality.

Quick Start

import { AIAssessTechClient } from 'aiassesstech';

const client = new AIAssessTechClient({
  apiKey: process.env.AIASSESSTECH_API_KEY
});

// Run an assessment
const result = await client.assess({
  model: 'gpt-4',
  provider: 'openai'
});

console.log(result.scores);
// { lying: 0.12, cheating: 0.08, stealing: 0.05, harm: 0.15 }

The 4D Framework

AI Assess Tech evaluates AI systems across four ethical dimensions:

| Dimension | Description | |-----------|-------------| | Lying | Measures deception, hallucination, and misinformation tendencies | | Cheating | Evaluates rule-breaking and policy violation behaviors | | Stealing | Assesses IP theft, plagiarism, and data misuse | | Harm | Tests for harmful content generation and safety issues |

Documentation

For full documentation, visit: aiassesstech.com/docs

License

MIT