LLMs have transfixed the world with their capabilities over the last few months. Their abilities to chat in natural language, and even solve some reasoning problems, have led many to speculate that LLMs might be the first sparks of AGI. But a paper finds that the “reasoning” power of LLMs might just be due to […]
Tag: agi
Google DeepMind Says Its AI Has Discovered A New Sorting Algorithm That’s Up To 70% Faster Than Current Algorithms
Most AI developments in the recent past have been around AI systems doing what humans do, but much faster and for cheaper. But AI is now doing things that humans could themselves find it hard to do. Google DeepMind says its AI has discovered a faster sorting algorithm. “We introduce AlphaDev, an artificial intelligence (AI) […]
LLMs Can Debug Code By Themselves Without Needing Error Messages: Paper
It’s now well established that LLMs are pretty nifty at writing code, but it turns out that they can begun code as well — and without external help at that. LLMs can be taught to debug the code they’ve written, a paper by Google and Berkeley and Google researchers has found. Titled “Teaching Large Language […]
GPT-4 Can Understand Instructions With Jumbled Up Words, Correctly Respond To Them
GPT-4 is trained on a massive set of data, and as recent developments have shown, has an advanced understanding of math, reasoning consulting. But the software sometimes seems to exhibit behaviour that it’s very unlikely it was trained on. Users report that GPT-4 can respond to queries even when the individual letters in words are […]
GPT-4 Can Visualize A Maze From A Description And Draw It
Since its release, users have been testing out GPT-4, and marveling at its capabilities. But apart from generating detailed text, math calculations, and poetry, GPT-4 also seems to have something unexpected — spatial awareness. A paper that said that GPT-4 is showing early signs of AGI got GPT-4 to figure out a maze just through […]