Search the archive

HONDA AND NISSAN TO MERGE, HONDA WILL TAKE THE LEAD

Nissan and Honda want to merge by 2026 to create the world's third-largest car company.

ELON MUSK'S XAI LANDS $6B IN NEW CASH TO FUEL AI AMBITIONS

xAI raised $6 billion in Series C, doubling its valuation to $45 billion. The company aims to accelerate infrastructure and product development, using models like Grok to enhance capabilities on X. xAI continues to expand amidst legal challenges with OpenAI, leveraging data from Musk's ventures and planning further fundraising next year.

SRIRAM KRISHNAN NAMED TRUMP'S SENIOR POLICY ADVISOR FOR AI

Sriram Krishnan, former a16z general partner, will serve as senior policy advisor for AI at the White House Office of Science and Technology Policy under Trump. He will coordinate AI policy across the government and work closely with David Sacks. Krishnan's background includes leadership roles at Microsoft, Twitter, and Facebook, and he has a strong relationship with Elon Musk.

OPENAI TRAINED O1 AND O3 TO 'THINK' ABOUT ITS SAFETY POLICY

OpenAI enhanced reasoning and safety features in its new AI model family, o3, using a novel "deliberative alignment" process. This method improves the alignment of AI model responses with OpenAI's safety values by referencing those values during inference without human-written data. Deliberative alignment, combined with advanced techniques like synthetic data and reinforcement learning, makes o3 OpenAI's safest model yet. It is set to release in 2025.

UNDERSTANDING MEDICAL DECISION-MAKING

MedDec is a dataset that helps improve the extraction of medical decisions from clinical notes. It covers eleven different diseases. The dataset is annotated with ten types of medical decisions.

LLMS WILL ALWAYS HALLUCINATE, AND WE NEED TO LIVE WITH THIS

As large language models become more ubiquitous across domains, it becomes important to examine their inherent limitations critically. Hallucinations in language models are not just occasional errors, but an inevitable feature of these systems.

HIGH-SPEED ROBOTICS SIMULATOR

ManiSkill3 is an advanced, open-source robotics simulator designed for scalable learning and manipulation tasks.

A TOOLKIT FOR SPEECH EMOTION RECOGNITION

EmoBox is a comprehensive toolkit designed to address common issues in Speech Emotion Recognition (SER). It provides a multilingual, multi-corpus benchmark for both intra-corpus and cross-corpus settings, enabling easier comparison and reproduction of SER models.

CANCER DETECTION

SAM-Swin is a model for detecting laryngo-pharyngeal cancer (LPC) that uses advanced features from the Segment Anything Model 2 (SAM2).

HOW TO GET REAL GPU UTILIZATION METRICS

Nvidia-smi shows a measure of GPU utilization but it is the amount of time where at least one kernel is running, not a full measure of GPU usage. This work by Stas shows how you can get actual FLOP usage.

SO MANY TOKENS, SO LITTLE TIME: INTRODUCING A FASTER, MORE FLEXIBLE BYTE-PAIR TOKENIZER

GitHub has released a new open-source byte-pair tokenizer that improves speed and flexibility for large language models like Copilot, scaling better with linear complexity. This tokenizer addresses limitations of existing BPE algorithms by efficiently supporting dynamic token counts for real-time text operations. It outperforms popular libraries like tiktoken and Hugging Face in benchmarks, significantly boosting performance for various applications.

SCALING LAWS – O1 PRO ARCHITECTURE, REASONING TRAINING INFRASTRUCTURE, ORION AND CLAUDE 3.5 OPUS “FAILURES”

There's skepticism around AI scaling laws, citing challenges like data exhaustion and hardware limits, but major AI players like Amazon, Meta, and OpenAI are continuing massive investments in data center and custom silicon capabilities, signaling their faith in ongoing scaling potential. New paradigms beyond pre-training, such as synthetic data, Reinforcement Learning, and advanced fine-tuning, are helping overcome traditional scaling barriers. OpenAI's o1 release demonstrates that increasing test-time compute, leveraging multi-datacenter training, and exploring novel scaling dimensions can significantly enhance AI model capabilities.

HOW CLAUDE USES AI TO IDENTIFY NEW THREATS

Anthropic's Clio tool identified a coordinated effort to generate SEO spam using its chatbot, Claude, leading to the termination of the spammers' access. Clio uses machine learning to detect emerging threats and highlight unusual chatbot usage, assisting Anthropic's trust and safety team. The company encourages other AI labs to adopt similar monitoring strategies to address potential harms while exploring diverse user applications.

GOOGLE'S NEW JULES AI AGENT WILL HELP DEVELOPERS FIX BUGGY CODE

Google Jules is an AI code agent designed to fix coding errors for Python and JavaScript in GitHub workflows.

MICROSOFT RELEASES PHI-4 LANGUAGE MODEL TRAINED MAINLY ON SYNTHETIC DATA

Microsoft's new open-source language model, Phi-4, excels in solving math problems, outperforming even larger models like GPT-4o and Llama 3.3.

CHATGPT'S NEW PROJECTS FEATURE CAN ORGANIZE YOUR AI CLUTTER

OpenAI's new Projects feature for ChatGPT enhances interaction organization by grouping related chats and files within a named Project.

🎉 _Special Holiday Offer_ for TLDR readers! Try for $1

APPLE JOINS GOOGLE ANTITRUST FIGHT TO PROTECT $20B SEARCH PARTNERSHIP

Apple is seeking to intervene in Google's US antitrust trial over online search, saying that Google can no longer adequately represent its interests in the case. The company has concerns over its multi-billion dollar revenue-sharing agreement with Google. Apple received an estimated $20 billion from Google in 2022 for making Google the default search engine on Safari. The trial could potentially force Google to divest its Chrome browser or Android operating system to restore competition in the online search market.

GOOGLE IS USING ANTHROPIC'S CLAUDE TO IMPROVE ITS GEMINI AI

Google contractors working to improve Gemini are comparing its answers against outputs produced by Anthropic's Claude. It is unknown whether Google has permission for its use of Claude in testing against Gemini. Anthropic's terms of service forbids customers from accessing Claude to build competing products or services or train competing AI models without approval. Google is a major investor in Anthropic.

NASA SPACECRAFT JUST PLUNGED INTO THE SUN AND BROKE STUNNING RECORDS

The Parker Solar Probe swooped through the Sun's atmosphere early on Christmas Eve. Flying around 3.8 million miles above the solar surface, the probe was designed to explore how solar storms are formed. It is fitted with instruments to measure and image solar winds, which may enable us to better forecast when and where a potent coronal mass ejection or solar flare may hit. Solar storms have huge implications for energy grids and communications systems on Earth.

UNITREE B2-W ROBOT DOG GETS UPGRADED, CAN NOW PERFORM A HANDSTAND

The Unitree B2-W robot dog can now do handstands thanks to a few upgrades, including high-powered wheel motors, M107 joint modules, and replaceable wheel/foot legs. It is equipped with a 32 wire automotive-grade LiDAR, a depth camera, and an ultra large 2200Wh battery, and can achieve a top speed of 12.4 miles per hour. It can travel 31 miles with an 88 pound load and pull up to 220 pounds. A video of the robot dog performing tricks is available in the article.

AUTHORTRAIL

AuthorTrail lets users explore Git repositories' history by author. It can find and browse files users have contributed to. Once found, users are able to quickly access to view and edit them. AuthorTrail features file search by author email, hierarchical file tree visualization, a built-in code viewer with syntax highlighting, quick open in the default editor, and dark mode support.

ASK HN: PROGRAMMERS WHO DON'T USE AUTOCOMPLETE/LSP, HOW DO YOU DO IT? (HACKER NEWS THREAD)

Many IDE features, like autocomplete, language servers, and Copilot, can be taken for granted today. However, not every developer uses these tools. Some developers work in environments where they aren't able to access IDEs (for example, on remote servers that only have Vim), so they have had to learn different ways of doing things. Organizing software in a way where every piece of code has a clear and logical place is a skill - tooling that helps with code navigation can result in code that lacks structure.

HOW TO (HOPEFULLY ETHICALLY) MAKE MONEY OFF OF AGI

This article contains an interview with Will Eden from Cosmos, Zvi Mowshowitz (an economist who spent 2.5 years trading at Jane Street Capital), and Noah Kreutter, who has roughly four years of quant-y finance experience at IMC and Bridgewater. The interview covers how to construct a portfolio that will go well when Artificial General Intelligence (AGI) becomes a bigger deal, the broad market effects of AGI, career capital in an AGI world, debt in interest effect of AGI, and much more. The interview was in November last year - the interviewer recently posted an update showing that the investment advice in the interview backtested well.

I SENSED ANXIETY AND FRUSTRATION AT NEURIPS'24

Late-year PhD students and postdocs at NeurIPS'24 expressed anxiety and frustration as the job market looks and feels very different from what they expected when they were applying for PhD programs five or so years ago. Years ago, competition for artificial intelligence researchers with PhDs was fierce - companies were recruiting them to prepare themselves for a revolution. Many of the researchers, who were paid well, were free to choose what they wanted to work on and publish whatever they wanted to publish. This incentivized a large number of brilliant students to apply to PhD programs. Universities have started to train undergraduate, graduate, and master's students on these technologies, making PhDs less valuable to companies.

DON'T LOOK NOW, BUT CHINA'S AI IS CATCHING UP FAST

Despite restrictions in buying advanced chips, China is showing signs of catching up with America's leading AI models more quickly than many in the industry had expected.

ASK HN: HAVE YOU EVER TAKEN A CAREER BREAK OR GAP YEAR TO HACK? (HACKER NEWS THREAD)

Taking a career break can be good for mental health and growth, but you have to weigh the decision against the opportunity cost of lost earnings, the amount of savings you have available, and your sense of value for the time you'll take off.

NETFLIX SUES BROADCOM'S VMWARE OVER US VIRTUAL MACHINE PATENTS

Netflix says that VMware's vSphere virtualization platform for deploying and managing virtual machines infringes on its patents related to virtual-machine communications.

APPLE'S BAD BLOOD WITH NVIDIA CONTINUES, AFTER DECADES OF FIGHTING

Apple is looking to further sever its ties with Nvidia by partnering with Broadcom to design its own AI server chip.

OPEN SOURCE MAINTAINERS ARE DROWNING IN JUNK BUG REPORTS WRITTEN BY AI

AI has ushered in a new era of slop security reports for open source.

OPENAI ‘CONSIDERED' BUILDING A HUMANOID ROBOT

OpenAI closed its robotics division in 2021 but it recently explored the topic of building its own humanoid robot again.

NHTSA FINALLY RELEASES NEW RULES FOR SELF-DRIVING CARS

The National Highway Traffic Safety Administration has introduced AV STEP, a framework aimed at facilitating the commercialization of autonomous vehicles while requiring companies to provide more data for public trust. The proposal seeks to ease exemption processes for driverless cars, potentially raising the cap on vehicles without traditional controls. However, its future is uncertain due to political and safety concerns.

SAUDI ARABIA INVESTS IN ROBOTS TO HELP BUILD ITS NEOM DESERT MEGACITY

Saudi Arabia's Neom has invested in Europe's GMT Robotics to enhance construction efficiency in its ambitious projects like the Line megacity. The focus is on automation and robotics to streamline rebar production and reduce human labor, given Neom's usage of 20% of the world's steel supply.

EX-TWITCH CEO EMMETT SHEAR IS FOUNDING AN AI STARTUP BACKED BY A16Z

Emmett Shear, former Twitch CEO, is launching an AI startup called Stem AI that will focus on AI that aligns with human behavior and ethics. It has already secured investment from a16z. Co-founded with Adam Goldstein, the startup is currently in stealth mode, but public records show active development since mid-2023. Shear, vocal on AI regulation and safety, has previously expressed concerns about AI's potential risks to humanity.

FASTER IMAGE & VIDEO GENERATION

Skip-DiT is a new version of Diffusion Transformers (DiT) designed to address the computational challenges in image and video generation.

GOOGLE'S TAKE ON AI AND CHESS

Google is celebrating chess' enduring influence on AI with global events and experiences that honor the game's impact on technology and creativity.

BAMBA EFFICIENT HYBRID MODEL (HUGGING FACE HUB)

Bamba is an efficient hybrid Mamba 2-style model with strong performance.

GEMINI LIVE MULTIMODAL API

Google has a prebuilt application that uses its new extremely fast multimodal API.

VISUALIZING 6D PARALLELISM

An amazingly detailed post exploring different training parallelism strategies for training deep models.

SCORALDET-DATASET

A dataset for detecting and classifying underwater coral species designed to facilitate marine conservation efforts using advanced AI models.

CHATGPT NOW UNDERSTANDS REAL-TIME VIDEO, SEVEN MONTHS AFTER OPENAI FIRST DEMOED IT

OpenAI released real-time video capabilities for ChatGPT, allowing users to interact with objects in near real-time using the Advanced Voice Mode with vision. The rollout is staggered, with some users like Enterprise subscribers receiving access in January. Google's similar feature, Project Astra, was also recently launched for trusted testers on Android.

ANTHROPIC'S 3.5 HAIKU MODEL COMES TO CLAUDE USERS

Anthropic has launched Claude 3.5 Haiku on its chatbot platform. It outperforms its predecessor, Claude 3 Opus, in coding and content moderation benchmarks. The model can generate longer text but lacks image analysis. Pricing controversies arose due to Anthropic's decision to increase the API cost after initially promising to match Claude 3 Haiku's pricing.

WHY MATERIALS SCIENCE IS KEY TO UNLOCKING THE NEXT FRONTIER OF AI DEVELOPMENT

The evolution from Intel's 1971 microprocessor to Apple's M2 Ultra highlights rapid semiconductor advancements driven by Moore's Law. However, as physical limits are reached, new materials and chip architectures like photonic and neuromorphic computing are crucial to support AI and next-gen technologies. The industry's future relies on innovative materials science to overcome scalability and energy challenges.

GOOGLE TOLD FTC MICROSOFT'S OPENAI DEAL IS KILLING AI COMPETITION

Google has urged the FTC to end Microsoft's exclusive cloud deal with OpenAI, arguing it raises costs for competitors.

GOOGLE'S NEW TRILLIUM AI CHIP DELIVERS 4X SPEED AND POWERS GEMINI 2.0

Google announced Trillium, its latest AI accelerator chip, boasting a 4x performance boost and significant energy efficiency improvements.

OPENAI INTRODUCES “SANTA MODE” TO CHATGPT FOR HO-HO-HO VOICE CHATS

OpenAI has added Santa Claus' voice to ChatGPT's Advanced Voice Mode for the holiday season - it is accessible via the snowflake icon across mobile and web platforms.

How does it work?

1️⃣ Set your departure airports.

XAI IS TESTING A STANDALONE IOS APP FOR ITS GROK CHATBOT

xAI is currently testing a standalone iOS app for its Grok chatbot in Australia and a few other countries. The app can access real-time data from the web and X. xAI is also preparing a dedicated site to make the chatbot accessible on the web. Grok was previously only available to X's paying subscribers, but the company rolled it out to all users for free earlier this month.