tag:blog.richardstu.com,2013:/posts Richards Tu's Blog 2025-03-09T13:58:37Z Richards Tu tag:blog.richardstu.com,2013:Post/2173480 2025-02-06T17:22:56Z 2025-03-09T13:58:37Z Looking Ahead to 2025

Although it's February now, I still think it would be quite nice to wrap up 2024 together with the beginning of 2025 and look forward into the new year.

The past thirteen months have been great, with a lot of good things happening. We've made huge progress across models' multimodal, reasoning, and agentic abilities, which are all important components on my own imaginary roadmap to capable AI systems that would have a huge impact on our species (or what people call "AGI").

]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2144825 2024-10-12T10:33:36Z 2025-01-08T17:44:31Z Envisioning Our Future with AI

This blog is inspired by Dario’s “Machines of Loving Grace”

]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2138394 2024-09-15T06:30:01Z 2024-12-28T16:21:44Z My Few Thoughts on OpenAI's o1 family models

Thinking Models Are Good Models

]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2135350 2024-08-19T08:00:00Z 2025-02-20T19:11:55Z My Few Thoughts on LLM's Reasoning Ability
People are having debate on this topic on X these days. Some ppl say that LLM can definitely reason because it can help us do math and code on some extent; but some other guys argue that LLM can't reason and they are not designed for it, what they do is just recite in training data.
]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2130300 2024-08-12T18:00:00Z 2025-01-08T04:32:48Z My Few Thoughts on Agents and Model's Autonomous Behavior
This is definitely one of the hottest topic these days lol. And I personally think that except the agent itself, its autonomous behavior is also really intriguing, since they're both related to how capable the base model is and also how dangerous the model can be.
]]> Richards Tu tag:blog.richardstu.com,2013:Post/2135348 2024-08-08T12:00:00Z 2025-02-20T18:56:31Z My Few Thoughts on Compute Scaling
To scale, or not to scale? It's a really interesting topic. Scaling Laws is a famous law in AI and ML, and the people's opinion's on it is also diverse. So I wanna share some thoughts on compute scaling, which is a part of the Scaling Laws.
]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2122567 2024-07-10T14:50:00Z 2024-12-28T16:26:55Z My Few Thoughts on AI Ethics

Tl;dr

  • AI development is rapidly outpacing societal adaptation
  • We should focus on adapting AI development to benefit human society
  • Concerns about job displacement can be mitigated by creating new AI-related roles
  • Combining AI with various scientific domains could accelerate breakthroughs
  • AI raises existential and philosophical questions about human uniqueness and purpose
]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2117815 2024-06-20T15:24:05Z 2024-06-21T08:18:55Z Testing with new Claude-3.5 Sonnet

Just now, Anthropic released their Claude-3.5 Sonnet (see the announcement here), and promised to release rest of the models in the model family later this year (i hate this 🫠).

Tl;dr

  • 3.5 Sonnet is more capable than 3 Opus yet cheaper;
  • 3.5 Sonnet is really good at reasoning tasks (it will ask you for "elaboration" frequently);
  • 3.5 Sonnet is good at vision;
  • 3.5 Sonnet has an updated knowledge cutoff date;
  • and more...

Current AI Development Path

Accelerate faster please.🥺

]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2116783 2024-06-16T13:34:58Z 2025-03-08T04:23:58Z Solution Sharing and Some Thoughts about Alibaba Global Mathematical Competition for AI

In Alibaba Global Mathematical Competition for AI, I created a system called "Self-Iterative Agent System for Complex Problem Solving"; however, I was surprised that it was so simple yet achieved such good results compared to others who used complex multi-agent systems. And for the details of my solution, I've uploaded to GitHub, you can find the URL in Read More. Feel free to share your opinion! 🤗

]]>
Richards Tu
tag:blog.richardstu.com,2013:Post/2116781 2024-06-06T10:38:00Z 2025-02-20T19:14:29Z My Few Thoughts on AI Security and AGI

Tl;dr

  • AI security crucial for future development; advocates for precautions
  • Seeks balance between AI capabilities and safety measures
  • AGI definition debatable; suggests exceeding average human ability in most domains
  • Emphasizes importance of AI understanding context-dependent language
  • Considers implications of advanced AI and hopes for beneficial outcomes

]]> Richards Tu