The second generation of Apple's Vision Pro headset is at least 18 months away from launch. Apple is keenly interested in feedback from customers who choose to return the first-generation headset as it will use the information to perfect the next version of the device. The Vision Pro will likely follow a refresh cycle longer than that of the iPad. There are rumors that Apple may launch a more affordable Vision headset around 2025.
ChatGPT has seen declining web traffic in five of the past eight months. It is currently down 11% from its May 2023 peak. Its mobile app has seen fewer users than Snapchat added last quarter alone. The competition in the AI space means that OpenAI will have to pump out hits to stay ahead.
The rate of information production is increasing faster than the storage density of tape. Using current technologies will mean that, in the coming decades, we will need exponentially more magnetic tape, disk drives, and flash memory - as well as the factories to produce the storage media. DNA is cheap, readily available, and stable at room temperature for millennia. Organizations around the world are already taking the first steps towards building a DNA drive that can both read and write DNA data.
MIT researchers have developed an antitampering ID tag that is tiny, cheap, and secure. It is several times smaller and significantly cheaper than the traditional radio frequency tags that are used to verify product authenticity. The tags use glue containing microscopic metal particles. This glue forms unique patterns that can be detected using terahertz waves. The system uses AI to compare glue patterns and calculate their similarity. The tags could be used to authenticate items too small for traditional RFIDs.
htmz is a minimalist HTML microframework for creating dynamic web user interfaces. It allows developers to load HTML resources within any element in a page. htmz only requires an inline HTML snippet to work - there are no dependencies and no backend is required.
This Reddit user recently put two complex codebases into GPT-4-Turbo-128K and Gemini 1.5 and asked them questions. The codebases implemented a parallel inet runtime, so it involved some hard compiler stuff. Gemini 1.5 completely outperformed GPT-4-Turbo-128K in the task of understanding the codebases. Most of the questions that GPT-4 got wrong are ones it would have answered correctly in a smaller context. Gemini still struggled to create a complete mental model of the system, but it is extremely good at locating existing information, making long-range connections, and doing some limited reasoning on top of this information.
Groq can serve up to 500 tokens per second. It is able to do this because it uses custom hardware that utilizes Linear Processor Units (LPUs) instead of GPUs. LPUs are designed to deliver deterministic performance for AI computations. They offer a more streamlined approach that eliminates the need for complex scheduling hardware, allowing every clock cycle to be utilized effectively. The system ensures consistent latency and throughput. LPUs can be linked together without the traditional bottlenecks found in GPU clusters, making them extremely scalable.
The Cocoa Press is a 3D printer that prints chocolate. It is now available for $3,995 prebuilt after a decade of development. Premade chocolate cores cost $49 per pack of 10. The printer allows users to program the heat of its nozzle to a tenth of a degree. It can create a wide range of shapes and textures, but there are still some problems to work out. This article shares what it is like using the machine. Pictures of chocolate prints made by the machine are available.