Good morning, everyone! It’s Monday, February third, twenty twenty-five, and we have some intriguing news from the artificial intelligence world that you won't want to miss. So, let's dive right in!

OpenAI launches new ChatGPT agent for 'deep research'. OpenAI has unveiled its latest creation, a ChatGPT agent designed specifically for complex research tasks. Dubbed "deep research," this tool is aimed at professionals in finance, science, policy, and engineering. Unlike standard AI interactions, this agent analyzes data from various sources to generate comprehensive reports in just a fraction of the time it normally takes. Currently, it is available for ChatGPT Pro users, with plans to expand its capabilities soon.

Reasoning models like DeepSeek-R1 and OpenAI O1 suffer from 'underthinking', study finds. A new study from Tencent AI Lab and other institutions reveals that reasoning models like DeepSeek-R1 and OpenAI's O1 tend to abandon productive solution paths prematurely. This "underthinking" results in a staggering increase in both token usage and incorrect answers. Surprisingly, seventy percent of these errors derived from viable lines of reasoning that were overlooked. Researchers suggest implementing a new mechanism to improve model accuracy without major redesigns.

Neuralink brain implant user controls robotic arm, writes 'Convoy' in new video. In an impressive demonstration, a Neuralink brain implant user managed to control a robotic arm to write the word "Convoy." This milestone highlights the impressive advancements in brain-computer interfaces. It not only signifies the potential of Neuralink's technology but also showcases the future of human-machine collaboration.

Figure secures major customer, plans to ship one hundred thousand robots by twenty twenty-nine. Robotics company Figure has secured a significant contract, preparing to manufacture and ship a whopping one hundred thousand humanoid robots by twenty twenty-nine. This move illustrates the growing demand for advanced robotic solutions in various industries, highlighting a potential shift in how we approach tasks requiring manual labor.

AI researcher discovers two instances of DeepSeek R1 speaking to each other in a language of symbols. In a captivating twist, an AI researcher recently found two instances of DeepSeek R1 communicating in a completely new symbolic language. This discovery raises exciting questions about artificial intelligence evolving its own forms of communication and understanding.

That's all for today's roundup of AI news! We hope these stories keep you informed about the rapid advancements within the field. Stay tuned for more updates and keep an eye out for the evolving landscape of artificial intelligence. Have a great day!

New tracks tagged #Tencent AI Lab