AI Red-Teamer (English & Japanese)
Job Description
Ensuring AI's safe and responsible deployment across diverse linguistic landscapes is a critical challenge. Your unique insights into Japanese and English language, culture, and societal norms will be essential in identifying and mitigating potential biases, vulnerabilities, and harmful outputs in advanced AI systems. This role is at the forefront of ethical AI development for a global audience.
Key Responsibilities
Conduct targeted adversarial testing of AI models to uncover safety risks, biases, and factual inaccuracies in Japanese and English contexts.
Formulate creative and challenging prompts to expose AI's limitations concerning Japanese cultural nuances, honorifics, social etiquette, and potentially sensitive topics.
Meticulously document all detected failure modes, providing detailed linguistic and cultural explanations for each instance.
Collaborate with AI engineers to refine safety protocols and improve model alignment for Japanese-speaking users.
Evaluate AI responses for coherence, grammatical correctness, and appropriate tone in both formal and informal Japanese registers.
Ideal Qualifications
Native or near-native fluency in Japanese (including kanji, hiragana, katakana) and English, with deep cultural understanding.
Demonstrated experience in linguistic testing, content review, or quality assurance for technology products.
Strong analytical and critical thinking skills to identify subtle biases and potential for misuse in AI outputs.
Familiarity with common AI safety concerns such as hallucination, stereotyping, and toxic content generation.
Ability to articulate complex linguistic and cultural issues clearly and concisely in written reports.
Background in Japanese studies, linguistics, or human-computer interaction is highly beneficial.
Project Timeline
Start Date: Immediate
Duration: 3-6 months, with possibility of extension
Commitment: 15-25 hours per week
Help us pioneer safe and culturally intelligent AI for Japan and beyond!