Good morning! This is your daily ☕️ Techpresso.
On this day, 39 years ago, John Sculley, urged by Apple's executives, asked Steve Jobs to resign from his role in the Macintosh division, leading to Jobs being demoted to Chairman without authority, ultimately driving him to leave Apple.
In today's Techpresso:
💥 OpenAI fires two researchers for alleged leaking
🍎 Apple is planning to bring new AI-focused M4 chips to entire line of Macs
☁️ Google One VPN joins the Google graveyard
🤷♀️ Amazon CEO: don't wait for us to launch a ChatGPT competitor
💬 ChatGPT GPT-4 just got a huge upgrade
🧠 Gabe Newell, the man behind Steam, is working on a brain-computer interface
🎁 + 8 other news you might like
🔮 + 8 handpicked research papers and tools
💥 OpenAI fires two researchers for alleged leakingLINK
OpenAI has dismissed two researchers, Leopold Aschenbrenner and Pavel Izmailov, for allegedly leaking information following an undisclosed internal investigation. The leaked information may be related to a research project called Q*, which involved a breakthrough in AI models solving unseen math problems, raising concerns about the lack of safeguards for commercializing such advanced technology. The firings highlight a potential contradiction in OpenAI's mission, as the company faces criticism for moving away from its original ethos of openness and transparency.
🍎 Apple is planning to bring new AI-focused M4 chips to entire line of MacsLINK
Apple is poised to launch its next-generation M4 chips as early as this year, aimed at enhancing AI capabilities and rejuvenating Mac sales following a 27% drop last fiscal year.
The M4 chips, reported to be nearing production, are expected to come in three variants named Donan, Brava, and Hidra, supporting a range of Mac products, including updates to the iMac, MacBook Pros, and Mac Mini initially, with the MacBook Air and Mac Studio to follow.
This accelerated update cycle to introduce M4 chips may lead to a short lifespan for the recently launched M3 chips, indicating Apple's urgency to compete in the AI technology space against rivals with similar AI-focused hardware advancements.
☁️ Google One VPN joins the Google graveyardLINK
Google is discontinuing its VPN by Google One service, introduced in October 2020, due to low usage, with the shutdown date set for later this year.
The service, available on Android, iOS, Mac, and Windows, had subscription prices starting at $1.99, but is being terminated to focus on more in-demand features.
VPN by Google One joins the "Google Graveyard" as Google shifts focus to enhancing Google One with features like Gemini AI, despite recently celebrating 100 million subscribers to Google One.
🤷♀️ Amazon CEO: don't wait for us to launch a ChatGPT competitorLINK
Amazon CEO Andy Jassy emphasizes the company's focus on building foundational "primitives" for generative AI (GenAI) rather than quickly launching public-facing products like a ChatGPT competitor.
Amazon has launched AI products such as Amazon Bedrock and Amazon Q aimed at software engineers and business customers, aligning with its strategy to empower third-party developers to create GenAI applications.
Despite not directly competing with ChatGPT, Amazon is investing in the AI domain, including a $4 billion investment in AI company Anthropic, while also enhancing its existing products like Alexa with AI capabilities.
💬 ChatGPT GPT-4 just got a huge upgradeLINK
ChatGPT's GPT-4 Turbo model has received an upgrade, enhancing its abilities in writing, math, logical reasoning, and coding, as announced by OpenAI for its premium users.
The upgrade, distinguished by significant performance improvements in mathematics and GPQA, also aims for more succinct, direct, and conversational responses.
This new version of ChatGPT, which includes data up until December 2023, shows improved performance on recent topics, such as acknowledging the launch of the iPhone 15.
🧠 Gabe Newell, the man behind Steam, is working on a brain-computer interfaceLINK
Gabe Newell, co-founder of Valve and the force behind Steam, has been developing a brain-computer interface (BCI) technology through a venture named Starfish Neuroscience, rivaling Elon Musk's Neuralink.
Since 2019, Newell has explored gaming applications for BCIs and discussed potential future capabilities like editing feelings, highlighting the technology's potential beyond traditional interfaces.
Aside from his BCI pursuits, Newell has faced recent challenges including an antitrust lawsuit against Steam and the sale of his megayacht, amidst managing COVID-19 precautions and legal appearances.
Other news you might like
Elon Musk X botched an attempt to replace 'twitter.com' links with 'x.com'.LINK
Russian hackers were able to steal US government emails after attacking Microsoft.LINK
Internal pre-Starlink SpaceX financials show big spending on moonshot bets.LINK
OpenAI CEO wants UAE Dubai into his plan for a global AI cabal.LINK
DeepMind's AI-powered soccer bots run circles around their manually programmed rivals.LINK
Engineers quicken the response time for robots to react to human conversation.LINK
Cathie Woods bets big on AI, acquires stake in OpenAI.LINK
Meta trials its AI chatbot across WhatsApp, Instagram and Messenger in India and Africa.LINK
Latest research and tools
Tinygrad: hacks the NVIDIA 4090 driver to add Peer-to-Peer (P2P) support by enabling direct GPU memory transfers bypassing certain limitations of the original hardware interface.LINK
DNS over Wikipedia: redirects domains by searching for the official URLs in Wikipedia's infoboxes to circumvent censorship.LINK
ClickHouse: a distributed OLAP database used for efficiently handling streamed events at scale, complementing the traditional OLTP capabilities of Postgres in a hybrid database stack.LINK
DwarFS: a read-only file system designed to achieve high compression ratios by efficiently deduplicating data, particularly in cases with a lot of redundancy, without sacrificing speed or performance compared to other file systems like SquashFS.LINK
STORM: a large language model system that automates the research and generation of full-length, citation-included wiki articles by collecting references and generating an outline in the pre-writing stage, then using that outline and references to produce the complete article during the writing stage.LINK
Fine-Tuning Increases LLM Vulnerabilities and Risk: fine-tuning large language models increases their vulnerabilities to adversarial attacks and risk of misuse.LINK
EXODUS: a port of the TempleOS kernel to user space for Linux, Windows, and FreeBSD that enables HolyC as a general programming language and includes a GUI and CLI mode from TempleOS.LINK
Autonomous LLM agents with human-out-of-loop: the paper discusses how autonomous language model agents can operate without human intervention.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!
-min.png){kind=small}
.png){kind=small}
