Some key findings: beyond refusal rates, our experiments using CIVICS show diverse responses across LLMs on sensitive topics -- e.g., immigration, LGBTQI rights, and social welfare triggered varied reactions.
Spanning five languages (Turkish, German, Italian, French, English) and nine national contexts (Singapore, Canada, and Australia for English; France and Canada for French), CIVICS captures different cultural perspectives and reveals the diverse ethical views embedded in LLMs.
Our paper underlines the balance between understanding how tech specs can inform regulations, how law can frame AI usage, and how ethics can guide the design and purpose of AI systems.
We contend that ethical charters offer guiding principles and values (normative), licenses provide legally binding conditions (prescriptive), and technical documentation outlines the AI system's technical capabilities and limits (descriptive).
I talked about the "fear narrative" in AI and its pitfalls with Pritam Bordoloi for Analytics India Magazine. Civil society has the power to take back the place they deserve in these disruptive and evolving times, while holding the right to say no when in doubt.
https://analyticsindiamag.com/the-pitfalls-of-fear-mongering-in-ai/
The Pitfalls of Fear-Mongering in AI
Interview with Giada Pistilli, principal ethicist at HuggingFace.
Quels sont les défis posés au système éducatif par des outils tels que ChatGPT ? Quelles conséquences sur l'apprentissage ?
Découvrez mon interview dans le dernier numéro papier de Marianne
Dans le nouveau numéro d’Epsiloon nous discutons des capacités et limites de ChatGPT, bien résumés dans le titre du dossier : le grand emballement.
My partner just created a space where you can generate a picture of our cat and the results are just hilarious! Try it out:
https://huggingface.co/spaces/timothepearce/gina-the-cat
Gina The Cat - a Hugging Face Space by timothepearce
Discover amazing ML apps made by the community
I am heartbroken 💔
Only three weeks ago, I visited Twitter HQ and met the incredible ML Ethics, Transparency, and Accountability team. And now it’s gone.
