Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Google releases its own 'reasoning' AI model | TechCrunch

Dec 19, 2024 - techcrunch.com
Google has introduced a new experimental AI model called Gemini 2.0 Flash Thinking Experimental, designed for reasoning and multimodal understanding in fields like programming, math, and physics. Available on AI Studio, this model aims to improve reasoning by using thoughts to strengthen its processes, although it is still in the early stages and shows room for improvement. Unlike traditional AI models, reasoning models like Gemini 2.0 can effectively fact-check themselves, reducing common AI errors but requiring more time to generate responses.

The release of Gemini 2.0 Flash Thinking Experimental follows a trend of reasoning models emerging from various AI labs, driven by the need for novel approaches in refining generative AI. Despite their potential, reasoning models are expensive due to the significant computing power they require and face skepticism regarding their long-term progress. Google is heavily investing in this technology, with over 200 researchers working on reasoning models, as the industry seeks alternatives to traditional "brute force" scaling techniques.

Key takeaways:

```html
  • Google has introduced a new experimental AI model called Gemini 2.0 Flash Thinking Experimental, designed for multimodal understanding, reasoning, and coding.
  • The model is part of Google's AI Studio and is intended to improve reasoning capabilities by using thoughts to strengthen its problem-solving processes.
  • Reasoning models like Gemini 2.0 Flash Thinking Experimental are gaining popularity, but they require significant computational resources and take longer to produce results.
  • There is debate over the effectiveness and future potential of reasoning models, as they are costly and their long-term progress is uncertain.
```
View Full Article

Comments (0)

Be the first to comment!