TLDR 2024-05-29
GPT-5 training starts š¤, Google Search leaks š, avoiding HTTPS redirects šØāš»
OpenAI training its next major AI model, forms new safety committee (2 minute read)
OpenAI has announced the formation of a new Safety and Security Committee to oversee risk management for its projects and operations. The company recently began training its next frontier model. The new Safety and Security Committee will be responsible for making recommendations about AI safety to the full company board of directors. It will be responsible for processes and safeguards related to alignment research, protecting children, upholding election integrity, assessing societal impacts, and implementing security measures.
An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them (32 minute read)
API documentation from Google Search was recently leaked. The 2,500-page document contained 14,014 attributes that appear to come from Google's internal Content API Warehouse. While it doesn't show things like the weight of particular elements in the search ranking algorithm or prove which elements are used in the ranking systems, it shows incredible details about the data Google collects. This article looks at the leak and extracts some of the most interesting details, for example, how Google utilized cookie history, logged-in Chrome data, and pattern detection to fight manual and automated click spam.
š
Science & Futuristic Technology
Will Scaling Solve Robotics? (15 minute read)
Over 900 people attended last year's Conference on Robot Learning, which featured 11 workshops and almost 200 accepted papers. One of the largest debates at the event was whether training a large neural network on a very large data set was a feasible way to solve robotics. This post presents the different sides of the argument to deepen people's understanding of the debate. Scaling has worked for other similar fields. However, it is impractical as there isn't much robotics data available and there's no clear way to get it. Even if scaling works as well as in other fields, it likely still won't solve robotics.
Neuralink rival sets brain-chip record with 4,096 electrodes on human brain (2 minute read)
Precision Neuroscience claims to have set a new record for the number of neuron-tapping electrodes placed on a living human's brain at 4,096. The high density of electrodes allowed neuroscientists to map the activity of neurons at an unprecedented resolution during a surgery to remove a benign brain tumor. Precision's device sits on top of the brain and does not penetrate the brain at all. It consists of a film that contains 1,024 electrodes embedded in a lattice pattern. The device is modular, allowing for multiple films to be used at once - four devices were used during the surgery.
Ex-OpenAI board member reveals what led to Sam Altman's brief ousting (3 minute read)
Former OpenAI board member Helen Toner revealed new details about what led to CEO Sam Altman's ousting in November in a recent interview. The board had apparently learned about ChatGPT's release on Twitter. Toner says that Altman had lied to the board multiple times. Altman was reportedly withholding information and misrepresenting things that were happening in the company for years. His lies made it difficult for the board to make decisions, leading them to think that he wasn't the right person to lead the company to AGI.
Ask HN: What would you spend your time working on if you didn't need money? (Hacker News Thread)
This Hacker News thread discusses what developers would choose to do for their ideal job. The answers range from large, world-benefiting schemes to personal projects and just doing nothing. Some of the answers are from people who have retired or are about to retire.
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