Davologist

In out, in out, shake it all about.

[Arvind Narayanan @random_walker]

There is widespread frustration about the AI industry’s “automate everything" agenda. But what would it take to steer AI toward augmenting human capabilities instead? In a new essay, Gabriel Unger (@_gabrielunger) and I lay out a research agenda and plan for action. https://t.co/gWjFI5VKbz

Our thesis is that the reason why AI-for-automation currently dominates isn’t just economic incentives but also deeper cultural issues in computer science as a field. And that can be changed.

- We sketch four important research directions that are needed in order to build collaborative or augmentative AI. The automation paradigm largely ignores these technical challenges.
- We advocate for Human-Computer Interaction and Human-Centered AI to be elevated as a subfield in status, influence, and funding, and changing AI curricula to emphasize the role of the human and decrease the centrality of the pure-automation problem framing.
- We point out that there are a surprising number of efforts in economics, computer science, industry, and policy all trying to bring about a shift toward augmentation, but not enough collective awareness. Better connecting these movements would amplify their power.
- We think the automation agenda is struggling on its own terms. Compiling evidence of the limits of general-purpose agents for automation and loudly making the case will encourage more companies to take collaboration more seriously as a goal.
- We predict that once more workers experience this way of working and recognize that there are alternatives to the delegate-and-supervise pattern, it will generate more demand, creating a flywheel effect.
- Targeted policy levers can play a small but important role in amplifying both the supply and demand for pro-worker AI. We don’t need clumsy top-down mandates.

We're not against automation. But we think the goal of AI must be broader.

This is the start of a long-term project. I know there are far more like-minded efforts out there than we’re aware of. I’d love to hear about some of them.

[News from Google @NewsFromGoogle]

For the first time, scientists have mapped the complete brain and central nervous system of an adult male fruit fly — a key model organism in science. 🪰

Working alongside HHMI Janelia Research Campus and the scientific community, @GoogleResearch scientists and researchers used AI to combine millions of 2D images into 3D neural shapes, reconstructing a record-breaking 166,000+ neurons. This foundational map of the adult male fruit fly brain can help accelerate our understanding of the brain, and is a major milestone in neuroscience.

[Sean McClure @sean_a_mcclure]

Efficiency is fragility.

There’s no way around this tradeoff.

A thing works well because it is surrounded by causally opaque dependencies.

An efficiently designed process strips your work of these unseen pieces.

If you make it efficient, expect it to break.

If you have been efficient for a while without problem, expect it to break catastrophically.

[Primers Of Raphael Douady @DouadyPrimers]

The #Primers are now announced on LinkedIn.

Primer on Panesthetic Analysis: June 1-4, 2026
Primer on AI & Machine Learning: June 5-6, 2026

Venue: Institut Louis Bachelier, Palais Brongniart, Paris

Instructors: Raphael Douady, Patrice Kiener

Registration link in my bio.

[Primers Of Raphael Douady @DouadyPrimers]

Since many #RWRI alumni have already attended my Primers on probability, risk, dynamic processes, AI & ML, it's high time that I teach what I've been developing for ~20yrs:

Panesthetic Analysis & Polymodels

June 1-6 in Paris, registration opens soon

Don't miss early-bird rate

[Davologist @davesdyke]

@ishirubi Imagine looking at a deep night sky, thinking you know what it is, in all its deepness.

[Jared Zoneraich @imjaredz]

Great report by the team at @trychroma @kellyhongsn

These are my practical take-aways for prompt engineers (CONTEXT engineers):

Your 100k token prompt is making your model dumber. even on tasks as simple as "repeat this string" too many tokens seriously degrade performance (we knew that though..)

Stop dumping full chat history. 113k token conversation history is causing 30% accuracy drops. Be surgical - extract only what matters for THIS query.

Models get distracted. the more similar info (but wrong), the more your model gets sidetracked or confused. Aggressively dedupe

Blobs work better than coherent structure. models get lost following structure when navigating long context. Counter-intuitively, non-repetitive (CONFUSING) chunks seem to work better. It's like when you skim an essay that's repetitive

TAKEAWAY: the million-token context window is a lie.

engineer your context like you're paying $ per token. in terms of accuracy you are

[Nick O'Brien @NickLovesSpain]

Spain has the 4th highest number of UNESCO World Heritage Sites —

But more World Heritage Cities than any country on Earth.

Here’s the list of 18 — the dazzling, iconic, and slightly unexpected:

(a 🧵)

[Davologist @davesdyke]

@Breedlove22 Or go here https://t.co/NQzsuQ3Tkn

[Davologist @davesdyke]

@mathladyhazel They’re everywhere 😍

[AK @_akhaliq]

CroissantLLM 🥐

A Truly Bilingual French-English Language Model

paper page: https://t.co/ARDUN4XcKq

introduce CroissantLLM, a 1.3B language model pretrained on a set of 3T English and French tokens, to bring to the research and industrial community a high-performance, fully open-sourced bilingual model that runs swiftly on consumer-grade local hardware. To that end, we pioneer the approach of training an intrinsically bilingual model with a 1:1 English-to-French pretraining data ratio, a custom tokenizer, and bilingual finetuning datasets. We release the training dataset, notably containing a French split with manually curated, high-quality, and varied data sources. To assess performance outside of English, we craft a novel benchmark, FrenchBench, consisting of an array of classification and generation tasks, covering various orthogonal aspects of model performance in the French Language. Additionally, rooted in transparency and to foster further Large Language Model research, we release codebases, and dozens of checkpoints across various model sizes, training data distributions, and training steps, as well as fine-tuned Chat models, and strong translation models. We evaluate our model through the FMTI framework, and validate 81 % of the transparency criteria, far beyond the scores of even most open initiatives. This work enriches the NLP landscape, breaking away from previous English-centric work in order to strengthen our understanding of multilinguality in language models.

[Prophetic @PropheticAI]

INTRODUCING MORPHEUS-1

The world’s first multi-modal generative ultrasonic transformer designed to induce and stabilize lucid dreams.

Available for beta users Spring 2024

[prerat @prerat]

asking ai to generate a 90s website but i keep asking it to be more 90s

[Davologist @davesdyke]

“It’s funny how the cards fall in a family - the actors change, but the roles stay the same.”

[Davologist @davesdyke]

@tsaralounga Border line, I’d say

[Nassim Nicholas Taleb @nntaleb]

SIC ET NON (version 2)

For technology; against neomania
For scholarship; against academism
For science; against scientism
For heuristics; against fortunecookism
For depositor bailout; against Geithnerism (saving bankers)
For central banking; against Bernankism
For skepticism; against nihilism
For risk-taking; against incaution
For government; against local overreach
For government; against bureaucrats
For youth; against glorifying ignorance
For down-to-earthism; against philistinism
For intervention; against interventionism
For defense; against warmongerism
For localism; against autarchy
For centralization; against mis-scaling
For justice; against toxic wokism
For medicine; against overtreatment
For probability; against statisticophasterism
For prudence; against paranoia
For wealth; against luxurious sybaritism
For globalization; against monoculture
For nations; against nationalism
For defense; against militarism
For education; against scholasticism
For religion; against intolerance
For rent; against rent-seeking
For doubt; against conspirationism

[Tim Urban @waitbutwhy]

Imagine the Earth were one meter in diameter.

Everything toweringly high up in our world—airplane cruising altitude, the top of Mount Everest, almost all clouds—is less than a millimeter off the surface.

Everything that's incredibly deep down—the bottom of the deepest ocean trench, the very deepest man-made hole—is less than a millimeter beneath the surface.

Our entire vast world exists in a razor-thin spherical shell.

[SwiftOnSecurity @SwiftOnSecurity]

Troubleshooting walkthrough:

Tonight I need to write a narrative of a case where a user complained a new browser add-in broke their mouse. This got escalated to me as the final tier.

I'm going to lay it out here first, because saying I'm working while laying in bed sounds cool.

[Juraj Bednar🏴💛🌘 @jurbed]

This is horrible, but it pushes us to what I've been saying for some time now - we need to embrace Progressive web apps more. Deploy them from our own home nodes.

Twitter has a pretty good PWA, I removed native app and use the web one only. If we do this, we won't need any store's approval. No politics and tocs required.

The slides are from my presentation Cypherpunk visions and trends 2023-2025, which talks about the content from the ebook by the same name.

[Saloni @salonium]

You often hear that full siblings "share 50% of their DNA."

But did you know this is actually the median and the range is pretty wide, roughly between 40–60%? https://t.co/KLeu4KnwgP
https://t.co/sna006rqR9