Saturday, July 13, 2024

Techpresso logo

TECHPRESSO

Meta lifts Trump's social media restrictions

Receive a daily summary of what happened in tech, powered by ML and AI.

Thank you! We sent you a verification email.
Oops! Something went wrong while submitting the form.
Join 1,500+ thinkers, builders and investors.

Good morning! This is your daily โ˜•๏ธ Techpresso.



In today's Techpresso:

๐Ÿ’ฅ Meta lifts Trump's social media restrictions

๐Ÿ“ OpenAI working on secret 'Strawberry' project

๐Ÿš€ SpaceX rocket accident leaves Starlink satellites in wrong orbit

๐Ÿค” OpenAI reportedly 'squeezed' through safety testing for GPT-4 Omni in just one week

๐ŸŽจ Generative AI helps creativity but hurts originality, study says

๐ŸŽ + 4 other news you might like

๐Ÿ”ฎ + 3 handpicked research papers and tools

๐Ÿ’ฅ Meta lifts Trump's social media restrictionsLINK

  • Meta has lifted the additional penalties for rule violations on former President Donald Trump's Facebook and Instagram accounts, allowing him to be treated like other users.
  • The prior restrictions, which allowed for full suspensions over minor violations, were implemented due to "extreme and extraordinary circumstances," according to Meta's Nick Clegg.
  • Trump, who has 34 million Facebook followers and 25 million Instagram followers, has resumed frequent posting since June 30th.
  • ๐Ÿ“ OpenAI working on secret 'Strawberry' projectLINK

  • OpenAI is developing a new reasoning technology under the code name 'Strawberry,' aimed at advancing the reasoning abilities of its AI models, according to a source and internal documents seen by Reuters.
  • The project, which remains highly confidential, involves a specialized post-training process designed to enable AI to not only answer questions but also autonomously browse the internet and conduct advanced research.
  • Similar to a method called "Self-Taught Reasoner" from Stanford, Strawberry aims to accomplish long-horizon tasks requiring extensive planning and action, which could significantly enhance AI models' reasoning capabilities.
  • ๐Ÿš€ SpaceX rocket accident leaves Starlink satellites in wrong orbitLINK

  • SpaceX's Falcon 9 mission to deploy 20 Starlink internet satellites experienced an anomaly with its upper stage, leading to an issue during the engine restart to raise the satellite orbit.
  • Though the Starlink satellites were successfully launched, they were placed in a lower orbit than intended, and efforts to raise their altitude using ion thrusters failed due to high atmospheric drag.
  • The FAA has mandated an investigation into the anomaly to ensure public safety and determine the root cause, working closely with SpaceX to implement corrective measures.
  • ๐Ÿค” OpenAI reportedly 'squeezed' through safety testing for GPT-4 Omni in just one weekLINK

  • OpenAI reportedly condensed the safety testing duration for its newest AI model, GPT-4 Omni, to just one week.
  • Sources indicated internal criticism, suggesting the company prioritized rapid deployment over thorough risk assessment, with some feeling pressured to hasten protocols for the set May launch date.
  • An OpenAI representative acknowledged the compressed timeline but affirmed that extensive internal and external tests were conducted, and the company is reconsidering its approach to such launches.
  • ๐ŸŽจ Generative AI helps creativity but hurts originality, study saysLINK

  • The study found that AI tools helped individuals with low creativity write more original short stories but reduced the overall creativity of groups using AI.
  • AI provided less naturally creative participants with suggestions that enhanced their writing quality, while highly creative individuals saw little to no benefit from AI assistance.
  • Access to AI generated more similar stories among participants, raising concerns that widespread AI use in creative tasks might lead to a decline in collective originality.
  • Other news you might like

    OpenAI illegally barred staff from airing safety risks, whistleblowers say.LINK

    Nearly all AT&T subscribersโ€™ call records stolen in Snowflake cloud hack.LINK

    Linus Torvalds says RISC-V will make the same mistakes as Arm and x86.LINK

    Meta researchers distill System 2 thinking into LLMs, improving performance on complex reasoning.LINK

    Latest research and tools

    Common Expression Language interpreter written in Rust: provides a lightweight tool for evaluating expressions in CEL, a simple and safe language with syntax similar to C++, Go, Java, and TypeScript, ideal for resource-constrained environments.LINK

    gpu.cpp: a lightweight library that streamlines the integration of GPU computing into C++ projects, compatible across major GPU brands and platforms, with minimal dependencies and a focus on rapid development cycles.LINK

    Exploring the Limits of Transfer Learning with a Unified Transformer (2019): the paper demonstrates how a single Transformer model can be effectively applied to various language processing tasks with minimal task-specific adjustments.LINK


    Want to get the latest news differently? Find us on:

    twitter instagram instagram instagram


    See you tomorrow for a new dose of โ˜•๏ธ Techpresso!

    Feeling behind on AI?

    You're not alone. Techpresso is a daily tech newsletter that tracks the latest tech trends and tools you need to know. Join 150,000+ professionals from top companies like OpenAI, Apple, Google and more. 100% FREE.
    Thank you! We sent you a verification email.
    Oops! Something went wrong while submitting the form.
    Join 1,500+ thinkers, builders and investors.
    You're in! Thanks for subscribing to Techpresso :)
    Oops! Something went wrong while submitting the form.
    Join 5,000+ thinkers, builders and investors.