(I originally wrote that Gemini removed thinking traces. People in the comments pointed out that they are accessible from a menu, so I did a new tweet. But I cannot believe how useless the thinking traces are. It makes Gemini outputs completely unauditable and thus untrustworthy)
ETHICS
-
Critique: Gemini hides its thinking traces and provenance
By
–
Google has hidden thinking traces on the Gemini site. You have to use the 3 dot menu to pull up summaries, which are so minimal as to be unusable. Did it do web searches? Did it check results? You can't tell. This makes Gemini unsuitable for any serious work you need correct.
-
Study: Human persuasion increases LLM compliance to objectionable requests
By
–
Our paper is out in PNAS: we found classic human persuasion techniques worked on AIs in a "parahuman" way, making them agree to objectionable requests (upping compliance from 35% to 51%) It worked on a range of major LLMs though newer models resist more https://
pnas.org/doi/10.1073/pn
as.2535868123
… -
Agentic engineering raises stakes in AI development
By
–
Nick Jablonski at #Rev26: vibe coding lowered the barrier. Agentic engineering raises the stakes. Knowing when humans must stay in the loop, and how to design for it, is the new engineering challenge. #AgenticAI #Rev26
-
Practical Guide to Reinforcement Learning from Human Feedback (RLHF)
By
–
New release from @PacktDataML at http://
amzn.to/3PMn1ZL "A Practical Guide to Reinforcement Learning from Human Feedback (RLHF)" 𝗔𝗺𝗮𝘇𝗼𝗻 𝘀𝘂𝗺𝗺𝗮𝗿𝘆: RLHF is a powerful approach to AI alignment and human-centered machine learning. By combining reinforcement learning -
Impact and risks of LLM-driven scientific productivity
By
–
At the same time concerns about uncritical adoption, need for guard rails “There is growing evidence that LLM-driven increases in scientific productivity often come at a cost.”
https://
nature.com/articles/d4158
6-026-01557-x
… -
AI Alignment and the Transfer of Human Values
By
–
If humans don’t intrinsically value other humans, then their machines will have even less of an incentive to do so. They are literally learning from us right now, absorbing our values and attitudes.
-
The physical and energy constraints limiting universal AI adoption
By
–
Une adoption universelle de l'#IA (au même titre que l'électricité ou l'eau potable) est totalement impossible en pratique.
— Olivier Rimmel 🕊️ (@OlivierRimmel) 19 mai 2026
Voici les quatre murs infranchissables qui bloquent cette vision.
1. Le mur de l'énergie et des réseaux électriques
L'IA générative est un gouffre… https://t.co/snCR2eYAZ0Une adoption universelle de l'#IA (au même titre que l'électricité ou l'eau potable) est totalement impossible en pratique. Voici les quatre murs infranchissables qui bloquent cette vision. 1. Le mur de l'énergie et des réseaux électriques
L'IA générative est un gouffre -
EU AI regulation critique and preemptive harm concerns
By
–
I don't believe pre agreeing anything works – eg EU did AI regulation which they started drafting before ChatGPT, as a result it is pretty much harmful without any of the benefits. Preparing some framework before we know what the world looks like could be harmful too. For
-
The Ethical Imperative of AI Governance and Human Oversight
By
–
La inteligencia artificial es una herramienta muy poderosa, pero si nadie nos enseña a usarla bien puede ser muy peligrosa. NO podemos dejar el destino de la humanidad en manos de un algoritmo… pic.twitter.com/fAot5wRZR4
— Juan Merodio (@juanmerodio) 19 mai 2026Artificial intelligence is a very powerful tool, but if no one teaches us how to use it well, it can be very dangerous. We CANNOT leave the fate of humanity in the hands of an algorithm…