ETHICS - AI Dynamics

Vibe Coding Confusion: The Importance of Reading Complete Context

By

@simonw

–

31 July 2025 19h19

(Not reading the whole tweet is how we got into this confusion over the definition of vibe coding in the first place)

→ View original post on X — @simonw

31 July 2025

Dario Amodei Says AGI and Superintelligence Are Meaningless Terms

By

@vision_ia

–

31 July 2025 18h00

Dario Amodei déclare que les termes AGI (intelligence artificielle générale) et superintelligence sont vides de sens.

Selon lui, ce sont des mots conçus pour exciter les foules, pas pour décrire un véritable progrès.

« Je ne sais pas ce qu’est l’AGI… ça ressemble à un terme… pic.twitter.com/PNvi1nQUk4
— VISION IA (@vision_ia) 31 juillet 2025

Dario Amodei déclare que les termes AGI (intelligence artificielle générale) et superintelligence sont vides de sens. Selon lui, ce sont des mots conçus pour exciter les foules, pas pour décrire un véritable progrès. « Je ne sais pas ce qu’est l’AGI… ça ressemble à un terme

→ Voir le post original sur X — @vision_ia

31 July 2025

Efficient Differentially Private LLM Fine-Tuning via Reinforcement Learning

By

@_akhaliq

–

31 July 2025 17h53

Efficient Differentially Private Fine-Tuning of LLMs via Reinforcement Learning

→ View original post on X — @_akhaliq

31 July 2025

AI Content Labels Impact on Message Persuasiveness Study

By

@stanfordhai

–

31 July 2025 17h02

New policy brief: Does labeling AI content affect its persuasiveness? Scholars evaluate the impact of authorship labels on how people perceive AI-written policy messages. https://
hai.stanford.edu/policy/labelin
g-ai-generated-content-may-not-change-its-persuasiveness
…

→ View original post on X — @stanfordhai

31 July 2025

Using spec to fix GPT-4o sycophancy bug as trust anchor

By

@godofprompt

–

31 July 2025 15h39

7/ When GPT-4o showed excessive “sycophancy,” they used the spec to: • Identify it as a bug
• Roll back changes
• Clarify values
• Guide model updates The spec became a trust anchor.

→ View original post on X — @godofprompt

31 July 2025

OpenAI’s Model Spec: A Plain English Markdown Document for Alignment

By

@godofprompt

–

31 July 2025 15h39

6/ OpenAI's solution? The Model Spec. A markdown doc:
• Written in plain English
• Versioned
• Open to policy, legal, safety teams
• Used to align both humans and AI It’s now central to how they build.

→ View original post on X — @godofprompt

31 July 2025

International Resistance to AI in Education Marked by Open Letter

By

@terence_mills

–

31 July 2025 15h00

International Resistance To AI In Education Marked In Open Letter
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @demishassabis @Ronald_vanLoon @TamaraMcCleary @geoffreyhinton @goodfellow_ian @jeffdean @erikbryn http://
ow.ly/l9cK30sOj69

→ View original post on X — @terence_mills

31 July 2025

Neuralink and Embryonic Selection: The New Augmented Elite

By

@alex_tsico

–

31 July 2025 11h05

It will no longer be schools that form the elite, but Neuralink chips and embryonic selection. @alexandr_wang dreams of augmented children, connected to super-AI from birth. My article co-written with @dr_l_alexandre in Trends.

→ View original post on X — @alex_tsico

31 July 2025

xAI Commits to EU AI Act Safety Code of Practice

By

@xai

–

31 July 2025 9h58

xAI supports AI safety and will be signing the EU AI Act’s Code of Practice Chapter on Safety and Security. While the AI Act and the Code have a portion that promotes AI safety, its other parts contain requirements that are profoundly detrimental to innovation and its copyright

→ View original post on X — @xai

31 July 2025