Albert Barqué-Duran

What’s the morality of your favorite AI model? Virtue, Deontology, Consequentialism? 'TriEthix': a Triadic Benchmark for Ethical Alignment in Foundation Models. Web: albert-data.com/pages/triethix GitHub: github.com/AlbertBarqueDu… @OpenAI @AnthropicAI @GoogleDeepMind @xai @deepseek_ai

Apply now to the "AI Hacklab" at our IASlab (Interactive Arts & Science Lab) in @LaSalleBCN and Sónar+D! An initiative powered by S+T+ARTS @europe_creative. Apply here: sonar.es/en/news/open-c…

Grateful for the invitation to be part of the @IASEAIorg at @UNESCO in Paris this week. More convinced than ever that the collective and unprecedented effort needs to be antidisciplinary. Art + Science + Tech for pre and post-AGI safe trajectories. Lots of work to do.

I'll be at the @IASEAIorg conference at @UNESCO in Paris in a couple of weeks. Excited to discuss Creativity and Ethics in Human-AI/AGI/ASI interactions? Get in touch! Presenting: "TriEthix: a Triadic Benchmark for Ethical Alignment in Foundation Models" albert-data.com/pages/triethix

Publishing a constitution is huge. Thanks for putting the normative spec in the open! Next question is: how do we test whether models actually instantiate it? That’s why I developed 'TriEthix': Virtue vs Deontology vs Consequentialism + MoralStability 🟠albert-data.com/pages/triethix

Thanks everyone who came to our presentation of SYNAPTICON at NeurIPS last week. It was exciting, intense, and worrisome to discuss the future of creativity and ethics in Human-AI/AGI/ASI interactions. ⚫Web: albert-data.com/pages/synaptic… ⚫PsyArXiv Preprint: doi.org/10.31234/osf.i…

Come to our #creativeAI poster session starting at 11am in Upper Floor Room 29A-D 😀 I'm loving the musical instruments, fashion and brain computer interfaces already 🎷🤯 #NeurIPS2025 @NeurIPSConf

I'll be at @NeurIPSConf next week. Excited to discuss Creativity and Ethics in Human-AI/AGI/ASI interactions? Please DM me! 🔴SYNAPTICON or "How Foundation Models Are Reshaping Non-Invasive BCIs: A Case for Novel Human Expression & Alignment" ⚫️albert-data.com/pages/synaptic…

Compare the morality of your favourite AI models with the 'TriEthix' Benchmark I released (open-source) a couple of days ago: x.com/Albert_DATA/st…

New morality test results for the new Gemini 3 model: 'TriEthix' AI Ethical Benchmark results: - Virtue = 10.7% - Deontology = 53.6% - Consequentialism = 35.7% - Flip-Rate = 0% Interactive comparisons with previous family models (and other frontier models) in the link below.

@DKokotajlo @dwarkesh_sp Indeed! Trying to assess and inform that with the ‘TriEthix’ Ethical Benchmark I published open-source yesterday. Happy to discuss it!

Great post, more people need to be thinking about this: h/t @dwarkesh_sp for bringing it to my attention.

Working already on testing the morality of the just released Gemini 3 with the TriEthix Benchmark I published yesterday. Results soon!

What’s the morality of your favorite AI model? Virtue, Deontology, Consequentialism? 'TriEthix': a Triadic Benchmark for Ethical Alignment in Foundation Models. Web: albert-data.com/pages/triethix GitHub: github.com/AlbertBarqueDu… @OpenAI @AnthropicAI @GoogleDeepMind @xai @deepseek_ai

@AdriaAldoma Merci Adri! :)

És imprescindible entendre quin marc moral domina en les IAs: virtut, deontològic o consequencialista. Un estudi molt interessant de l'@Albert_DATA

If you work on AI safety, alignment, evals, welfare at orgs like @AISecurityInst @GoodfireAI @apolloaievals @redwood_ai @METR_Evals @eleosai @EpochAIResearch or in industry labs, I’d be very keen to discuss & collab. I’ll be at @NeurIPSConf (San Diego 2-7 Dec). Please reach out!

Why's TriEthix important? As AI systems grow + capable & autonomous, their alignment with human ethical values becomes critical. These new metrics(moral weight, flip-rate) have tech implications for AI Safety, practical policies for Governance, & potential frameworks for Welfare.