iiinova — where your digital self keeps evolving

Alberto Hernández Marcos @alberto-h.bsky.social · 19/07/2025

Apple M1 Pro con Sequoia 15.5 va como una seda

Alberto Hernández Marcos @alberto-h.bsky.social · 19/07/2025

Creo que nos ha pasado a todos los que hacemos investigación. Lo he resuelto (casi todo ) pidiendo con firmeza en el prompt que compruebe las fuentes e incluya URL/DOIs, y habilitando el Web search antes de que genere la respuesta. Perplexity.ai es una gran alternativa al de siempre, por cierto.

perplexity.ai

Perplexity

Perplexity is a free AI-powered answer engine that provides accurate, trusted, and real-time answers to any question.

020

Alberto Hernández Marcos @alberto-h.bsky.social · 19/07/2025

Gracias por aportar valor a esta plataforma. La realidad es que estar fuera de Twitter/X es pagar un precio muy alto todavía.

020

Reposted by Alberto Hernández Marcos

César Astudillo @cesarastudillo.bsky.social · 19/07/2025

Este artículo me ha parecido muy provocador y me ha abierto muchas dudas. Quizá un feminismo plural y dialogante pueda tener conversaciones incómodas y constructivas con una fracción de los hombres objetivo del discurso manosférico, pero temo que para los propios manosféricos ya sea tarde para eso

151

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 15/06/2025

A big AI question is why, as LLMs get bigger, their values seem to increasingly converge on the same preferences, this holds for Musk’s Grok & China’s DeepSeek, too. “These findings suggest that value systems emerge in LLMs in a meaningful sense, with broad implications” arxiv.org/abs/2502.08640

2418722

Reposted by Alberto Hernández Marcos

Carl T. Bergstrom @carlbergstrom.com · 09/06/2025

Ok, time for a short thread about this paper. My sense over the past six months or so is that chain-of-thought prompting as used in e.g. ChatGPT o.3 improves substantially upon previous systems such as ChatGPT 4.o, at least for certain tasks. But how revolutionary is it?

1628892

Reposted by Alberto Hernández Marcos

Dean Baker @deanbaker13.bsky.social · 27/05/2025

I strongly second Krugman's "letter to Europe." The EU should tell Trump to take his tariff and shove it. Hitting U.S. consumers with a huge tax increase is not smart policy, but as a reality TV show star, what does Trump know about economics? paulkrugman.substack.com/p/a-letter-t...

paulkrugman.substack.com

A Letter to Europe

You’re stronger than you think. Act like it.

17685180

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 22/05/2025

Individuals keep self-reporting huge gains in productivity from AI & controlled experiments in many industries keep finding these boosts are real, yet most firms are not seeing big effects. Why? Because gaining from AI requires organizational innovation. www.oneusefulthing.org/p/making-ai-...

oneusefulthing.org

Making AI Work: Leadership, Lab, and Crowd

A formula for AI in companies

25512

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 20/05/2025

Big: The final version of a randomized, controlled World Bank study finds using a GPT-4 tutor with teacher guidance in a six week afterschool program in Nigeria had "more than twice the effect of some of the most effective interventions in education" ("equating to 1.5 to 2 years" of standard school)

1216833

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 16/05/2025

I wish these skeptical AI articles (this is from the NYTimes) would actually grapple with the growing body of research that AI can do original research & perform key unstructured tasks across the spectrum of high-end white collar employment. AI criticism is important, but it should be clear-eyed.

5527

Alberto Hernández Marcos @alberto-h.bsky.social · 16/05/2025

Awesome. Where is this taken from? Thx

100

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 10/05/2025

A common question is "can an AI make money?" This benchmark, where AIs run a simulated vending machine over time, suggests yes, with an important caveat On average, Claude 3.5 & o3-mini beat a human, but are high in variance & fail at random times for complex reasons. andonlabs.com/evals/vendin...

86615

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 08/05/2025

"Our findings demonstrate that reasoning models improve not only the clarity, organization, and professionalism of legal work but also the depth & rigor of legal analysis itself." Law students using o1-preview had the quality of their work on most tasks increase (up to 28%) & time savings of 12-28%

1545

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 06/05/2025

I just don’t see signs of a major increase in hallucination rates for recent models, or for reasoners overall, in the data. It seems like some models do better than others, but many of the recent models have the lowest hallucination rates.

38012

Reposted by Alberto Hernández Marcos

Melanie Mitchell @melaniemitchell.bsky.social · 06/05/2025

I'm looking forward to this event next week in Amsterdam!

1253

Reposted by Alberto Hernández Marcos

ruggsea @ruggsea.eurosky.social · 07/05/2025

Reposting from the other site to spread it here: apparently, thinking that RLHF irons out creativity in LLMs is now corroborated by this paper arxiv.org/pdf/2505.00047

1173

Reposted by Alberto Hernández Marcos

Keezy Young🌼 @keezyyoung.bsky.social · 07/05/2025

the specific laughter of a little kid who is being taken on a ride of some kind (tricycle, plastic car, thrown up in the air by a dad etc) is one of the most precious and valuable things you'll ever hear

11568

Reposted by Alberto Hernández Marcos

Melanie Mitchell @melaniemitchell.bsky.social · 02/05/2025

Karpathy: We have reached "jagged Intelligence" Mollick: We have reached "jagged AGI" Next up: "jagged consciousness"? www.oneusefulthing.org/p/on-jagged-...

oneusefulthing.org

On Jagged AGI: o3, Gemini 2.5, and everything after

New models and new thresholds

6426

Reposted by Alberto Hernández Marcos

César Astudillo @cesarastudillo.bsky.social · 29/04/2025

A quienes teníais pensado acudir: por desgracia habrá que aplazar el evento porque la UCM ha suspendido todas las actividades. Se establecerá una nueva fecha a corto plazo y por supuesto os la contaré.

041

Reposted by Alberto Hernández Marcos

Gus @gusthema.bsky.social · 28/04/2025

Gemma 3 are just amazing models! but what if you want to manipulate it's internal activations to understand how it does its text generation? Sascha Rothe is here to teach you how! Great insights for anyone curious about the inner workings of LLMs! www.youtube.com/watch?v=JTUs...

youtube.com

Inside Gemma 3: Modifying the output through activation hacking

YouTube video by Google for Developers

1104

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 28/04/2025

👀Today’s AIs are already hyper persuasive. A controversial study where LLMs tried to persuade users on Reddit found that: “Notably, all our treatments surpass human performance substantially, achieving persuasive rates between three and six times higher than the human baseline.”

911015

Reposted by Alberto Hernández Marcos

Píxel Sonoro @pixelsonoro.bsky.social · 28/04/2025

🎵🎶¡Mañana celebramos este evento en homenaje a la obra de @cesarastudillo.bsky.social y @riskwood.bsky.social donde profundizaremos en la creación musical durante la "edad de oro" ✅Exposiciones ✅Interpretación de arreglos orquestales de algunas de sus piezas ¿Cómo puedes seguirlo? ⬇️⬇️

155

Reposted by Alberto Hernández Marcos

Phillip Carter @phillipcarter.dev · 10/04/2025

I have yet to see an "intro to AI" video as comprehensive but also approachable as Andrej Karpathy's 1hr overview. I maintain that anyone who watches this (and pays attention the whole time) will come away with an intuitive understanding of how to use LLMs www.youtube.com/watch?v=zjkB...

youtube.com

[1hr Talk] Intro to Large Language Models

YouTube video by Andrej Karpathy

1172

Reposted by Alberto Hernández Marcos

Kate Beaton @katebeaton.bsky.social · 12/04/2025

Richard Serra lived seasonally in our village so we went to the Reina Sofia to see their room of his work, but the children had no love for grand abstract minimalist rectangles, and we were forced to depart after much whining, it was a quixotic quest anyway

A person standing in front of a Richard Serra artwork

82523

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 03/04/2025

If you wanted to see how little attention folks are paying to the possibility of AGI (however defined) no matter how much the labs publicly discuss it, here is an official course from Google Deepmind whose first session is "we are on a path to superhuman capabilities" It has less than 1,000 views.

2214416

Alberto Hernández Marcos @alberto-h.bsky.social · 02/04/2025

And this applies to so many other approaches, like building-my-own-RAG...

010

Reposted by Alberto Hernández Marcos

Serge Belongie @serge.belongie.com · 30/03/2025

Would you present your next NeurIPS paper in Europe instead of traveling to San Diego (US) if this was an option? Søren Hauberg (DTU) and I would love to hear the answer through this poll: (1/6)

docs.google.com

NeurIPS participation in Europe

We seek to understand if there is interest in being able to attend NeurIPS in Europe, i.e. without travelling to San Diego, US. In the following, assume that it is possible to present accepted papers ...

6279161

Reposted by Alberto Hernández Marcos

Pete Marcus @petemarcus.bsky.social · 27/03/2025

Useful chart to understand AI agents (by @mmitchell.bsky.social) www.technologyreview.com/2025/03/24/1...

0197

Reposted by Alberto Hernández Marcos

Paul Graham @paulgbot.bsky.social · 15/03/2025

043

Reposted by Alberto Hernández Marcos

Melanie Mitchell @melaniemitchell.bsky.social · 20/03/2025

In my latest column for Science magazine, I discuss recent AI "reasoning" models -- how it works, to what extent it captures "genuine" reasoning processes, and what's needed to answer such questions. www.science.org/doi/10.1126/...

science.org

Artificial intelligence learns to reason

Julia has two sisters and one brother. How many sisters does her brother Martin have?Solving this tiny puzzle requires a bit of thinking. You might mentally picture the family of three girls and one b...

715759

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 20/03/2025

This trend seems to be growing here. False comfort that AI doesn’t work or that it isn’t getting better is pervasive on Blue Sky. As a result, people who could add important points of view to current discussions on the meaning & use of AI instead try to believe they don’t have to think about it.

2224641

Reposted by Alberto Hernández Marcos

Ethan Mollick @emollick.bsky.social · 04/03/2025

Not to be a broken record, but AI critics who insist that AI "doesn't work" and is going to just disappear are misleading - that just isn't true, as controlled studies like this one show. There are many issues with AI & many things that need critique, but pretending it is going away is not helpful.

921732

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Is this the way to “give machines a heart”? Maybe! We’re excited to see where it leads. Drop your thoughts below or check out the paper for details. LOVE (Latest Observed Values Encoding): nature.com/articles/s41598-024-72817-x (This work was part of my PhD thesis at #UGR)

nature.com

A generic self-learning emotional framework for machines - Scientific Reports

Scientific Reports - A generic self-learning emotional framework for machines

010

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Will emotion increase AI agents’ utility? Boost learning? Maybe unlock social skills or interpretability? Applications in education, arts, entertainment, healthcare... the possibilities are vast. I feel like this is just the beginning of an exciting journey.

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

So what’s next? Well, now that AI can develop its own “limbic system”, how can it integrate into an agent’s behavior (or “mind”)? Here’s one way we could try:

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

This proved that humans could infer the emotional state of the agent just from observing its actions! LOVE was bridging the gap between internal AI states and emotions that made sense to us humans. 🤯

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

But were these synthetic emotions genuine? We validated LOVE with an emotional attribution study. In it, humans could consistently identify the agent’s emotions, closely matching what LOVE had self-learned. 📊🤖 We also compared them to five academic accounts. 📚✨

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Analytical comparison revealed matches with 7 basic emotions, one appearing in two intensity levels. And yes, during landing episodes, these emotions showed situationally coherent patterns: elicitation, decay, natural transitions, and even moments of confusion.

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

We needed “interpretability mappings” to put names to those patterns. Here’s the LOVE 2:5x6 mapping we used, as “ideal references”. Can you match the learned patterns to the 30 ideal references? 👇

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

So, would it actually learn “emotions” from the rewards and state-values observed by an agent? When tested on a classic RL case study (LunarLander-v2), it learned these 8 patterns. But what did they mean???

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Here’s an overview of the framework, including an emotional encoder (observations -> emotion vector) and an interpreter (emotion vector -> word). We used an autoencoder because they can learn great low-dimensional representations of complex patterns.

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Ok, so emotions = patterns in values... 🤔 And hey, ML is ALL about learning patterns, right? So why not learn those patterns from the sequences observed by the RL agent? That’s EXACTLY what we did.

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

The examples above are Order III LOVE patterns (with two values) of a few basic emotions as interpreted by the framework. Lower and higher orders are possible, suggesting a natural gradient to more complex emotions (but that’s another story)... 🐛…🐦…🐶…🧑🏽…❤️✨

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Alternative continuation: a competitor arrives first! After the initial surge, expectations drop... but not all the way down: the agent isn’t giving up on that sweet prize just yet! Yes, the agent would probably feel “anger” now...

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Now an opportunity arises, bringing a wave of “yummy” positive rewards. Expectations soar first, then gradually settle as rewards materialize. Would the agent feel... “excitement”, followed by “satisfaction”?

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Think of this struggling agent: low energy, scarce opportunities, every step costs a little (tiny negative rewards). The sequence for rewards (green) and expectations (orange, the value function) would look like Pattern 1. Would the agent feel... “concern”? (feel ≈ synthesize)

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

LOVE is built on a simple but powerful novel hypothesis: Emotions correspond to “distinct temporal patterns perceived in crucial values for a living being, such as recent rewards, expected future rewards or anticipated world states.” But how does that make sense?

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

So what’s LOVE? It’s a generic framework for RL agents to learn emotions through interactions with any environment without supervision. No preset labels, hard-coded dynamics or pre-engineered features. Think of it as “artificial emotional intelligence” for machines! 🤖 🧠✨

100

Alberto Hernández Marcos @alberto-h.bsky.social · 19/03/2025

Well, what we DON’T know is how exactly they emerge in our brains. Engineers don’t have an “emotion handbook” to implement them into a #ReinforcementLearning (RL) agent, for example. And that’s exactly the focus of this research. #AffectiveComputing #EmotionAI

100