Knowledge catalogue
arXiv:2605.31590v1 Announce Type: cross Abstract: Text-to-video (T2V) generation faces challenging questions when generating videos with long horizons containing multiple events. Inspired by the intri
arXiv:2605.30930v1 Announce Type: cross Abstract: As large language models (LLMs) increasingly act as collaborative partners, human--AI alignment is often evaluated through explicit task success, accu
arXiv:2505.18069v3 Announce Type: replace Abstract: Hebbian and anti-Hebbian plasticity are widely observed in the brain and are classically modeled as mechanistic, local homosynaptic rules stabilized
arXiv:2605.30576v1 Announce Type: new Abstract: Exploration in reinforcement learning for autonomous driving is inherently unsafe: agents must experience novel behaviors to learn, yet exploration can
arXiv:2506.22304v3 Announce Type: replace-cross Abstract: Continuous Normalizing Flows (CNFs) enable elegant generative modeling but remain bottlenecked by their iterative nature requiring costly samp
arXiv:2605.31521v1 Announce Type: new Abstract: Semantic speech tokenizers have become a widely used interface for Audio-LLMs, owing to their compact single-codebook design and strong linguistic align
arXiv:2605.31040v1 Announce Type: new Abstract: Developing effective representations for register transfer level (RTL) designs is crucial for accelerating the hardware design workflow. Existing approa
arXiv:2605.31131v1 Announce Type: cross Abstract: Attention-deficit/hyperactivity disorder (ADHD) is a psychiatric disorder which presents itself in individuals through patterns of developmentally ina
arXiv:2605.31524v1 Announce Type: new Abstract: Certification methods for stochastic systems provide sufficient proof rules, based on real-valued supermartingale certificates, to determine the almost-
arXiv:2605.30451v1 Announce Type: new Abstract: Group Relative Policy Optimization (GRPO) is an effective recipe for training reasoning models with verifier-based outcome rewards, but its supervision
arXiv:2605.31556v1 Announce Type: cross Abstract: Alignment teaches vision-language models (VLMs) to avoid expressing demographic biases, and when gender is clearly visible they largely succeed. Far l
arXiv:2605.30877v1 Announce Type: new Abstract: Large-scale Vision-Language-Action (VLA) pretraining is increasingly adopted as the foundation for robot policies, yet the evidence for pretrained VLAs
@webisticsdawg @GaryMarcus Personally I think everyone w a 401(k) should be absolutely livid. I'm disgusted by this, it's so brazen. How dare the richest man in the world pick working Americans' pocke
arXiv:2605.31561v1 Announce Type: new Abstract: Test-time reasoning has become a significant field of study since the introduction of chain-of-thought reasoning in large language models (LLMs). Howeve
What if the skeptics are right and superintelligence is impossible? Great! Then a proactive ban costs us nothing, prevents massive compute waste, and hurts no one. But if they are wrong? We face an un
arXiv:2605.30719v1 Announce Type: cross Abstract: We study when large language models (LLMs) can serve as effective black-box policy optimizers for reinforcement learning (RL) tasks, i.e., when can we
which is NOT new; see this quote from 5 years ago. the fact that all this is still true says a lot. This was right five years ago, and still is: “Large scale pretrained models are certainly likely to
arXiv:2506.00175v5 Announce Type: replace-cross Abstract: Modern AI systems are typically developed through multiple stages-pretraining, fine-tuning rounds, and subsequent adaptation or alignment, whe
arXiv:2605.31261v1 Announce Type: cross Abstract: The family of linear recurrent neural networks has shown strong performance as recurrent memory units in partially observable reinforcement learning.
arXiv:2604.01985v2 Announce Type: replace-cross Abstract: General-purpose world models promise scalable policy evaluation, optimization, and planning, yet achieving the required level of robustness re
arXiv:2603.10422v2 Announce Type: replace Abstract: World Models (WMs) offer a promising mechanism for post-training Vision-Language-Action (VLA) policies by providing dynamics priors that improve gen
arXiv:2605.30833v1 Announce Type: cross Abstract: On-policy distillation transfers reasoning capabilities by training a student model on its own generated trajectories using token-level feedback from
arXiv:2605.30612v1 Announce Type: cross Abstract: Continuous control policies trained with off-policy reinforcement learning frequently exhibit high-frequency action jitter, rendering direct deploymen
arXiv:2605.30896v1 Announce Type: new Abstract: Bidding in repeated auctions is a central challenge for reinforcement learning (RL), combining continuous control with the strategic complexities of dig
Anyone remember how I said in January 2025 that AI was going to be stumble and be dubbed “too big to fail”, along with cries for bailouts? If that call was correct – which increasingly seems likely, a
Gary Marcus discusses the Pope's understanding and perspective on artificial intelligence, examining statements or positions the religious leader has taken regarding AI technology and its implications
I am absolutely with @GaryMarcus and the Pope on this. (Not something you expect to say everyday). We are not creating beings. Systems don’t exp. grief nor hope, hold a value construct, or the ability
Listen, I’m a big “the index is the index” guy, but they are openly looting the coffers. This is 100% fraud. Rule changes for the SpaceX SPCX IPO: Index providers waived the profitability requirement
@ParValue26 @Hedgeye Let’s be clear what this actually is: the administration and the world’s richest man working together to screw over ordinary investors to goose returns on the IPO for the wealthie
serious accusation. does this fit with people’s experience? Is Anthropic altering model performance to force costly upgrades? Chapter Co-Founder and CEO @CobiBGantz outlines a shift his team recently
SpaceX being rammed into indices with no profit requirements, seasoning, and generally looser constraints is economic terrorism. Index trackers will eat the loss when reality catches up and retail inv
Gary Marcus argues that consumer backlash against AI-generated content will be increasingly visceral, particularly in creative fields like fiction, leading to growing demand for 'human-authored' certi
Three companies are about to IPO at a higher combined value than all 2,600 dot-com IPOs from 1995 to 2000 combined. – SpaceX, OpenAI, Anthropic (2026): ~3.75 trillion – Every dot-com IPO from 1995–200
Weird how the Pope seems to understand AI better than @geoffreyhinton, but I am 100% with the Pope on this. We are NOT creating beings. The Pope is right. We are creating interactive fiction that is t
weird the way this tweet was getting a ton of traffic and then just stopped. 🤷♂️ I honestly think Elon’s best days are behind him: BYD is crushing Tesla in EVs. Waymo is crushing Tesla in AVs. Anthro
This post likely critiques the irony of someone insulting another person's intelligence while making a basic grammatical error (omitting an apostrophe in 'don't'). Gary Marcus, a cognitive scientist a
genius reply, full of reasoned intellectual argument, from the kind of retail investor who is probably going to get burned on the SpaceX IPO. Going to MIT and NYU may have meant something in the past,
Gary Marcus discusses his moderation approach on social media, noting that he blocks Tesla supporters primarily when they resort to insults rather than substantive arguments, and observes that this oc
I honestly think Elon’s best days are behind him: BYD is crushing Tesla in EVs. Waymo is crushing Tesla in AVs. Anthropic, Openai, and Google are crushing Xai on AI. The SpaceX S-1 is so ridiculous th
“I'm more concerned about the lack of intellectual diversity within the frontier AI commentariat/research world. This improved a lot over the last two years, but we're still far from a healthy ecosyst
is having a four month lead a sustainable multitrillion dollar business model? We took another look at the capability gap between open-weight and proprietary models. Since the start of the year, open-
it is the single most clever move Elon ever pulled off Rule changes for the SpaceX SPCX IPO: Index providers waived the profitability requirement and cut the seasoning window from 90 days to 5. This f
More of my thoughts on the topic in @Corriere: https://www.corriere.it/esteri/26_maggio_29/yoshua-bengio-ai-pope-6aa95de2-33f7-4f73-8905-34b1abedbxlk.shtml “Like nuclear energy, AI must be at the serv
Gary Marcus expresses frustration about an influx of Elon Musk supporters on his social media feed who conflate personal insults with substantive counterarguments. The post appears to critique a lack
so much for multitrillion dollar candy companies Note to all staff: Turns out that super tasty AI candy we've been putting out in large bowls isn't free and it isn't necessarily leading to better prod
this won’t end well. it will end with a bailout. Cash flow no longer covers the AI capex bill, so hyperscalers are funding it with record debt: Hyperscaler bond issuance has soared to $150 billion YTD
what a time to be alive wow, Opus 4.8 is very... argument-happy? it picked a fight with me about my usage of the word 'ontology', and when we eventually got back on the same page philosophically, told
Where is the power/value of money physically located? It's clearly not in the actual physical bills. Why are we more scared of an elderly mafia boss than their much more physically dangerous underling
Why LLMs rarely payoff—and what I have been saying literally for 7 years—confirmed yet again: LLMs can’t handle the truth. (Nor apparently can my critics, who keep saying I am “always wrong”, when I h
A 25B fund just refused SpaceX at any price. It says the company can't be worth more than 1T, half the $1.8T IPO target, and Musk's 85% control makes it impossible to fix from inside. https://thenextw
arXiv:2605.29975v1 Announce Type: new Abstract: We present a fully convolutional denoising autoencoder (FC-DAE) for denoising two-time intensity-intensity correlation functions (C_2) in X-ray photon c
arXiv:2605.29673v1 Announce Type: cross Abstract: Reconstruction-based inference assigns a class by comparing class-wise reconstruction residuals; Sparse Representation Classification (SRC) is a canon
arXiv:2605.30313v1 Announce Type: new Abstract: Simulation-based RL for contemporary robot control is increasingly organized around GPU-resident simulation: physics, rollout collection, and learning a
arXiv:2605.28824v1 Announce Type: new Abstract: Constructing artificial lexicons that are pronounceable, typologically plausible, and semantically structured remains an open challenge in computational
arXiv:2605.30070v1 Announce Type: cross Abstract: Moving beyond simple scalar rewards toward richer world feedback is a natural path to more scalable RL post-training. On-policy self-distillation (OPS
arXiv:2605.29791v1 Announce Type: new Abstract: While Large Language Models (LLMs) can convincingly simulate personas in explicit self-reports, they often deviate in implicit behavioral decisions, rev
arXiv:2605.29458v1 Announce Type: cross Abstract: Accurately simulating the decisions of a specific individual remains challenging for large language models (LLMs), partly because persona information
arXiv:2603.01006v2 Announce Type: replace-cross Abstract: REPresentation Alignment (REPA) improves the training of generative flow models by aligning intermediate hidden states with pretrained teacher
arXiv:2605.28914v1 Announce Type: cross Abstract: Tool-using language agents turn model decisions into external side effects: they read files, run scripts, call APIs, send messages, and invoke Model C
arXiv:2605.29434v1 Announce Type: cross Abstract: Existing sentence-level watermarking methods enhance robustness to paraphrasing by anchoring watermarks in sentence semantics. However, their prefix-b