If you want to learn something about what AGI alignment researchers are studying, this is a fantastic post: https://astralcodexten.substack.com/p/deceptively-aligned-mesa-optimizers?s=r
Here is a video of what it looks like using DALL·E 2: https://twitter.com/karenxcheng/status/1513569345049280512
A striking mathematical idea could solve two hardware challenges in artificial intelligence: A neural network with a single neuron that is used many times may be as effective as large neural networks while using much less energy. https://thenextweb.com/news/how-ai-brain-with-only-one-neuron-could-surpass-humans
An unusual form of matter known as spin glass can potentially allow the implementation of neural network algorithms in hardware. https://phys.org/news/2022-03-circuits-rare-nanomagnets.html
OpenAI has extended GPT-3 to do research on the web when it needs information that it doesn’t already have. https://openai.com/blog/webgpt/
For the past 70 years, computing has been dominated by general-purpose hardware: machines designed to run any code. Even vector processors and their descendants (GPUs) are fundamentally general purpose. The next steps forward in AI may involve software, hardware, and neural networks that are designed for each other. https://www.ft.com/content/4db69e3c-c901-4776-970e-c57e99f71aba [archived version: https://archive.ph/XJ7EE]
Does this artificial intelligence think like a human? A new technique compares the reasoning of a machine-learning model to that of a human, so the user can see patterns in the model’s behavior. https://news.mit.edu/2022/does-this-artificial-intelligence-think-human-0406
Researchers have observed the process by which brains store sequences of memories. In addition to therapies for memory disorders, this discovery could lead to advances in artificial intelligence, which don’t really have the ability to create and process timelines or narratives. https://thenextweb.com/news/researchers-figured-out-how-the-human-brain-makes-memories
Dirty Pipe is a newly discovered high-severity bug in the Linux kernel that allows any user to overwrite any file or obtain root privileges. Android phones are also vulnerable. https://arstechnica.com/information-technology/2022/03/linux-has-been-bitten-by-its-most-high-severity-vulnerability-in-years/
Validate Production of Neutrons from Gas Gun-Driven Targets https://firstlightfusion.com/science-hub/validate-production-of-neutrons-from-gas-gun-driven-targets
Alexander Young: everything you want to know about cognitive genomics https://unsupervisedlearning.libsyn.com/alexander-young-everything-you-want-to-know-about-cognitive-genomics
Detailed piece by on the rising risk of lab-based outbreaks and why pandemic prevention research can inadvertently cause pandemics or be harnessed by bioterrorists. https://www.vox.com/22937531/virus-lab-safety-pandemic-prevention
There are now 8 times as many unicorns in France as when Macron took office https://www.france24.com/en/europe/20220117-france-celebrates-25th-unicorn-in-strong-start-to-2022-for-tech-sector
Most people don't realize how advanced the newest machine learning models are because of political correctness.
For example, last year, Facebook already unveiled a chatbot (Blender Bot 2.0) that is capable of "ongoing conversations that may continue for days, weeks, or even months." In other words, you could already talk to virtual assistants such as Alexa. The technology is there. But the companies are mortally afraid that these models might say something politically incorrect and create popular outrage.
The same is true for models that generate images. Companies are reluctant to release them because of the anticipated backlash if it generates an image of Muhammad or child porn.
The most important news source. It's a Kruel world.