Nvidia announces new AI chips months after latest launch as market competition heats up (1 minute read)
Nvidia has unveiled a new generation of artificial intelligence chip architecture called Rubin. The company only just announced its upcoming Blackwell model in March - those chips are still in production and expected to ship to customers later in 2024. Nvidia has pledged to release new AI chip models on a one-year rhythm. The less-than-three-month turnaround from Blackwell to Rubin underscores the competitive frenzy in the AI chip market.
|
SpaceX Aims to Launch Cellular Starlink Service This Fall (3 minute read)
SpaceX has revealed plans to launch its Starlink system for phones this fall in a filing about the FCC's new rules on supplying satellite connectivity to US carriers. The company says that the FCC's current framework for supplying satellite connectivity to phones is too restrictive and has urged the agency to loosen the aggregate limit on radio frequencies for cellular satellites, specifically the “one-size-fits-all aggregate out-of-band power flux-density.” It backs replacing the restriction with more granular band-specific limits. SpaceX's rivals have lobbied the Commission to maintain the restriction to protect against potential radio interference.
|
|
Science & Futuristic Technology
|
What would you do with a robotic third thumb? (6 minute read)
'The Third Thumb' is a 3D-printed robotic wearable designed to enhance users' biological capabilities. Controlled by users' toes, the robotic thumb sits on the side of the palm opposite users' biological thumbs, resembling a sixth elongated finger. It is operated through a pair of sensors under each big toe - pressure applied by the right toe moves the thumb side to side, while pressure from the left toe moves it up and down. The device can be used to complete tasks that would ordinarily be difficult or impossible to perform with just a thumb and fingers. A video demonstrating what The Third Thumb can do is available in the article.
|
AI is cracking a hard problem – giving computers a sense of smell (6 minute read)
Smell in humans is experienced through about 400 types of receptors in the nose. Scientists are using machine learning to help computers learn what certain molecules or sets of molecules smell like to humans. Machine learning is key to digitizing smells because it can learn to map the molecular structure of an odor-causing compound to textual odor descriptors. Dataset sizes have grown significantly in the past few years, allowing researchers to finally apply machine learning to machine olfaction. The technology has promising applications such as personalized perfumes and fragrances, better insect repellents, novel chemical sensors, early detection of diseases, and more realistic augmented reality experiences.
|
|
Programming, Design & Data Science
|
An interview with the most prolific jailbreaker of ChatGPT and other leading LLMs (13 minute read)
Pliny the Prompter posted a jailbreak for OpenAI's newest foundation model, GPT-4o, just a few hours after it was released. The jailbreak allowed users to get the model to output explicit copyrighted lyrics, instructions for making banned items, strategic plans for attacks, and medical advice based on X-rays. Pliny has been finding jailbreaks for leading large language models for around 9 months. This article contains an interview with the prompter where they discuss what motivates them, their approach to jailbreaking a new model, legal ramifications from AI jailbreaking, jailbreaking ethics, and more.
|
What We've Learned From A Year of Building with LLMs (45 minute read)
Building with AI is now more accessible than ever, but it can still be difficult to build products and systems that are effective beyond a demo. This guide teaches readers how to build successful products with large language models. It draws from the authors' experiences of building applications with LLMs over the past year. The guide discusses tactics, operations, and strategies for building with LLMs.
|
|
The Internet's Final Frontier: Remote Amazon Tribes (20 minute read)
SpaceX's Starlink has been making the internet accessible to previously isolated tribes in Brazil since 2022. This article looks at what happened to the Marubo tribe, a community that has preserved its way of life for hundreds of years through isolation, after it gained access to the internet. While everyone in the tribe was happy when the technology arrived as the internet brought clear benefits, such as the ability to chat with faraway loved ones and call for help in emergencies, people in the tribe feel things have gotten worse. The internet has become essential to the tribe, but at a cost.
|
The Art of Scaling Taste (24 minute read)
Startups can survive extremely competitive environments by having good taste. Taste is how a business differentiates itself when attention is scarce and choice is abundant. Knowing what to make is just as important as the ability to make it. MSCHF is a business that has managed to scale its taste without compromising on it. This article looks at how the company releases so many products while staying through to its original vision, how it keeps its team creative and engaged, and lessons from the company that can be applied by anyone.
|
|
‘Little Tech' sets its sights on Washington (12 minute read)
Garry Tan, the president and CEO of Y Combinator, is trying to build a lobbying operation that fights on behalf of Little Tech due to concerns that federal regulations meant to mitigate the potential harms of AI will instead help tech giants maintain their lead on the technology, boxing out smaller players.
|
Manifest (Website)
Manifest is a complete backend that features a database, admin panel, REST API, and JavaScript SDK that can be set up with just one file of simple code.
|
|
Love TLDR? Tell your friends and get rewards!
|
Share your referral link below with friends to get free TLDR swag!
|
|
Track your referrals here.
|
Want to advertise in TLDR? 📰
|
If your company is interested in reaching an audience of tech executives, decision-makers and engineers, you may want to advertise with us.
If you have any comments or feedback, just respond to this email!
Thanks for reading,
Dan Ni & Stephen Flanders
|
|
|
|