TLDR 2023-11-16
Google Search Notes 🔎, exploring GPTs 🤖, why people use VBA 👨💻
Microsoft is finally making custom chips — and they’re all about AI (8 minute read)
Microsoft built its own custom AI chip for training large language models. It has also developed Arm-based CPU for cloud workloads. Both chips will arrive in 2024. They could potentially lessen the industry’s reliance on Nvidia. Microsoft is already designing the second-generation versions of these chips. More details about the new chips are available in the article.
Google Search will let you leave and read ‘Notes’ on web pages (5 minute read)
Google is experimenting with a way for users to create and read Notes that contain tips from other people when searching for an article or topic. The feature is a response to the popularity of people adding ‘Reddit’ to their search queries for more human results. Users can like, share, save, or report Notes. More details about Notes and other new features Google recently added to Search are available in the article.
🚀
Science & Futuristic Technology
China Claims World’s Fastest Internet With 1.2 Terabit-Per-Second Network (1 minute read)
Huawei and China Mobile have built a 3,000-kilometer-long internet network linking Beijing to the south. The firms claim it is the world's first network to achieve a stable and reliable bandwidth of 1.2 terabits per second. The network uses domestically-owned key technologies.
Crispr gene editing shown to permanently lower high cholesterol (4 minute read)
A single infusion of a novel gene-editing treatment has been shown to reduce cholesterol in people with an inherited condition that causes extremely high LDL cholesterol levels. The treatment uses CRISPR to edit a gene in the liver. Gene editing could provide a longer-lasting option for treating hereditary high cholesterol compared to long-term medication. A lot of work still needs to be done to ensure that the treatment is safe.
💻
Programming, Design & Data Science
10 Best Practices for Securely Developing with AI (Sponsor)
Walk through some crucial tips and tricks to help you securely embrace AI technology and get some tips for protecting against AI-generated code risks such as prompt injection and data access.
MonkeyPatch (GitHub Repo)
MonkeyPatch is a method for easily calling an LLM in place of the function body in Python. It produces the same parameters and output you would expect from a function implemented by hand. MonkeyPatch produces functions that are well-typed, reliable, stateless, and production-ready.
The Architecture Of Serverless Data Systems (14 minute read)
The future of cloud data services is large-scale and multi-tenant. This article looks at real-world serverless multi-tenant (MT) data architectures to understand how different types of systems implement serverless MT. While some patterns immediately jump out, there is a surprising amount of diversity among the systems surveyed. The article covers what serverless MT is and the generic challenges in building these systems.
Amazon to staff: Come into the office – it'd be a shame if something happened to your promotion (3 minute read)
Amazon employees risk undermining their own promotion prospects if they don’t return to the office for three days a week. Corporate workers must get VP-level approval if they aren’t able to come into the office. Amazon had issued a return-to-office mandate in May. Around 10% of its employees protested the mandate, including those that joined the company during the pandemic as remote workers.
Exploring GPTs: ChatGPT in a trench coat? (33 minute read)
GPTs allows ChatGPT Plus users to create their own custom GPT chat bots that other Plus subscribers can then talk to. While it might not seem like much more than a fancy wrapper for standard GPT-4 with some pre-baked prompts, GPTs combines many features to provide much more than that. This article discusses these features and shows users how to get the most out of OpenAI’s latest feature.
Get the most interesting stories in startups, tech, and programming delivered in a free daily email.
Join 1,600,000 readers for
one daily email