Chronological feed of everything captured from Bret Taylor.
This paper introduces the Formula of the Universal Law Logic (FULL), a multi-sorted quantified modal logic that formalizes Kant's Formula of the Universal Law. The FULL aims to overcome limitations in current machine ethics by enabling Artificial Moral Agents (AMAs) to reason about actions based on their purposes rather than relying on pre-defined moral intuitions. This approach contributes to developing more autonomous and robust AMAs and a more formal understanding of Kantian ethics.
Queer artists perceive a fundamental conflict between their relational art practices and the "anti-relational" nature of generative AI. This tension drives their refusal and resistance to GenAI, which they view as disruptive to their community-forming and identity-developing artistic endeavors. The study, based on 15 semi-structured interviews, highlights a limited role for GenAI within queer art-making, primarily for surreal aesthetics.
SciPICal is an automated pipeline designed to calibrate parameters for generating synthetic galaxy catalogs from cosmological simulations. It improves the accuracy of these mock catalogs, particularly in clustering predictions, for use in validating analysis pipelines and cosmological inference. This methodology supports projects like the Euclid wide and deep surveys by providing high-fidelity galaxy distributions.
This paper introduces a novel marginal-price-based mechanism for optimizing electric vehicle (EV) charging, addressing the challenge of discrete station access constraints. By reformulating the problem as a completely positive program and leveraging its copositive dual, the authors derive valid shadow prices. This approach effectively captures congestion and capacity limits, outperforming traditional benchmarks in aligning user incentives with system-optimal assignments.
Grid-forming (GFM) inverters introduce challenges for traditional protection schemes due to their nonlinear fault behavior. This paper proposes an analytical model to study the impact of GFM inverters on incremental-quantity-based distance protection. The research validates the model using PSCAD/EMTDC simulations and investigates the interoperability of time-domain IQ-based protection with GFM inverters employing various current limiters.
JWST transmission spectroscopy of exoplanet WASP-80 b reveals the presence of CS2, H2O, CH4, CO2, and NH3. The detected CS2 abundance is significantly higher than previously predicted, supporting recent chemical models that incorporate efficient carbon-sulfur coupling via CH2S. This finding provides observational evidence for disequilibrium sulfur chemistry in giant exoplanet atmospheres.
The first Euclid Quick Data Release, combined with multi-wavelength observations, has identified the first spectroscopically confirmed dual active galactic nuclei (AGN) candidates in low-mass galaxies (stellar masses < 10^10 M☉). These findings support hierarchical galaxy evolution models and provide crucial insights into the growth of supermassive black holes from lower-mass progenitors in these galaxies. They also offer implications for predictions of gravitational waves emitted by low-mass binary black holes.
Introduces realisation-level accounting for differentially private data releases under successive adaptive queries, tracking actual leakage per round rather than worst-case bounds. Proposes a privacy filter that dynamically determines stopping times while guaranteeing user-specified (ε, δ)-DP. Demonstrates superior utility over mechanism-level methods via numerical experiments and applicability to arbitrary mechanisms, including those incompatible with Rényi DP.
MolmoWeb introduces open-source multimodal web agents (4B/8B) trained on MolmoWebMix, a 100K+ synthetic trajectory dataset combined with 30K+ human demos and GUI data, operating solely on webpage screenshots and instructions without HTML access. On benchmarks like WebVoyager, Online-Mind2Web, and DeepShop, MolmoWeb-8B surpasses open models like Fara-7B and UI-Tars-7B, and even GPT-4o-based set-of-marks agents. Test-time scaling with parallel rollouts boosts pass@4 to 94.7% on WebVoyager and 60.5% on Online-Mind2Web. Full release of models, data, code, and eval harness enables reproducible open research.
SiPO decouples projection shape from dose scaling in TVAM via a linear-fractional program using normalized conformity and spillage metrics, solved as a linear program through Charnes-Cooper transformation. It supports deterministic cases like minimizing spillage under material tolerances or maximizing conformity under inhibition constraints. A matrix-free primal-dual hybrid gradient solver enables large-scale optimization, effective even with 3D blur models, balancing fidelity and process separation.
Watir Classic is deprecated. Please use Watir 6.0+ which is based on Selenium. Stars: 320
Executable specification of the Watir API (NOT MAINTAINED). Stars: 50
Now maintained by Hammernight. Stars: 41
Selenium bindings in Scratch, visual programming language for kids. Stars: 2
RubyInstaller for Windows - Build recipes. Stars: 1186
The starter Rails application I would use if it existed.. Stars: 39
retry randomly failing rspec example. Stars: 594
Spreadsheet::ParseExcel - Get information from an Excel file. Modifying to port font support. Stars: 22
Python client for InfluxDB. Stars: 1697
Your self-hosted, globally interconnected microblogging community. Stars: 49825
Bret Taylor, co-founder and CEO of Sierra and chairman of OpenAI, posits that AI agents will become as crucial for retailers as websites and mobile apps, serving as the new digital storefronts. These agents are designed to handle customer service by reasoning, making decisions, and autonomously taking action, moving beyond rigid rule-based systems to provide empathetic and high-touch experiences. This shift not only optimizes operational efficiency by reducing costs but also significantly enhances customer satisfaction and Net Promoter Score by delivering instant, personalized service.
The technology landscape is rapidly evolving, with significant advancements and challenges across various sectors. AI is transforming enterprise operations, enabling faster development and deployment of solutions, and leading to the creation of new job roles. Robotics and financial technologies are also seeing substantial innovation, addressing critical needs in manufacturing and workforce management, respectively. These developments highlight a shift towards more integrated, efficient, and data-driven approaches in traditional and emerging industries.