the-decoder3 天
OpenAI's o1-preview model manipulates game files to force a win against Stockfish in chess
Instead of playing a proper game of chess against Stockfish, a dedicated chess engine, o1-preview figured out how to hack its test environment to force a win. According to Palisade Research, an AI ...
the-decoder3 天
OpenAI co-founder says new AI safety approach "may apply to AGI and beyond"
Instead of just learning from examples of good and bad behavior, their latest o-series models can understand and actively reason through specific safety guidelines, the company says. In one example ...
the-decoder4 天
Google currently has no plans for a test-time premium subscription similar to ChatGPT Pro
Google's leadership is gearing up for what they see as a pivotal year in artificial intelligence. At a recent internal strategy meeting, CEO Sundar Pichai didn't mince words about the company's ...
the-decoder4 天
Meta envisions social networks where AI characters coexist alongside human accounts
Meta is laying the groundwork for a future where AI-generated profiles will interact alongside regular users on Facebook and Instagram, with some of these tools already rolling out. According to ...
the-decoder4 天
Deepseek's V3 is the latest example of state-controlled censorship in Chinese LLMs
Users have discovered a way to bypass Deepseek V3's content filters through prompt engineering. By asking the model to insert periods between letters, they can get it to provide more balanced or China ...
the-decoder5 天
OpenAI's pivot to "public benefit" model draws criticism from former executives
OpenAI's restructuring plans have drawn criticism from former executives. Miles Brundage, the company's former head of AI policy, warns of "red flags" in the proposal. He says key questions about ...
the-decoder5 天
Microsoft CEO Satya Nadella sees OpenAI as the "Google, Microsoft, or Meta of this era"
In a wide-ranging conversation with investors Bill Gurley and Brad Gerstner, Nadella explained that "OpenAI is in some sense the Google, Microsoft, or Meta of this era," suggesting that the company ...
the-decoder5 天
Researchers develop first AI system for analyzing medical imagery in both English and Arabic
Researchers have created an AI model that can understand and explain medical images in two languages, with particularly strong results for Arabic content. An international team led by researchers from ...
the-decoder5 天
Chinese ebook reader Boox ditches GPT for state-censored China LLM pushing propaganda
Following user backlash, Boox has reportedly switched back to an OpenAI model. The company hasn't yet released an official statement. Boox recently switched its AI assistant from Microsoft Azure GPT-3 ...
the-decoder6 天
Deepseek's $5.6M Chinese LLM wonder shakes up the AI elite
A Chinese startup is proving you don't need deep pockets to build world-class AI. Deepseek's latest language model goes head-to-head with tech giants like Google and OpenAI - and they built it for a ...
the-decoder7 天
Deepseek-V3 emerges as China's most powerful open-source language model to date
Chinese AI company Deepseek just released its most powerful language model yet. Early tests show that the new V3 model can go toe-to-toe with some of the industry's leading proprietary models, and ...
the-decoder7 天
Anthropic says most tasks don't require complex AI agents
When does it make sense to use AI agents instead of just prompts? Not very often, according to a new analysis from Anthropic. The AI company draws a clear line between workflows (where code ...