Thursday, October 10, 2024
Advertisement
  1. You Are At:
  2. News
  3. Technology
  4. OpenAI's new model can solve complex math, physics problems with greater accuracy than many teachers

OpenAI's new model can solve complex math, physics problems with greater accuracy than many teachers

OpenAI’s GPT-4o was only able to solve 13 percent of the problems in the International Mathematics Olympiad correctly. The new model is more advanced and can think like humans before responding to a problem.

Written By: Om Gupta New Delhi Published on: September 14, 2024 11:20 IST
OpenAI new model
Image Source : FILE OpenAI new model

As generative AI (GenAI) gains popularity across various fields, OpenAI, led by Sam Altman, has unveiled a new 'reasoning' model. This model has been trained to answer complex questions faster than humans can. According to the AI company, the 'OpenAI o1 model' has been trained to take more time to think through problems before responding, similar to how a person would approach problem-solving. Through training, these models learn to refine their thinking process, explore different strategies, and recognize their mistakes.

The new AI model has various applications. It can be used by healthcare researchers to annotate cell sequencing data, by physicists to generate complex mathematical formulas for quantum optics, and by developers across different domains to build and execute multi-step workflows.

“We've developed a new series of AI models designed to spend more time thinking before they respond. They can reason through complex tasks and solve harder problems than previous models in science, coding, and math,” the company added.

During tests, the model performed similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology.

“We also found that it excels in maths and coding. In a qualifying exam for the International Mathematics Olympiad (IMO), GPT-4o correctly solved only 13 per cent of problems, while the reasoning model scored 83 per cent,” said OpenAI.

The coding abilities were evaluated in contests, and the model reached the 89th percentile in Codeforces competitions. As an early model, it currently lacks many of the features that make ChatGPT useful, such as web browsing for information, and uploading files and images.

However, for complex reasoning tasks, this marks a significant advancement and represents a new level of AI capability.

"Given this, we are resetting the counter back to 1 and naming this series OpenAI o1," the company announced.

OpenAI has also developed a more affordable model in the 'reasoning' series, called OpenAI o1-mini. This is a faster reasoning model that is particularly effective for coding tasks. As a smaller model, o1-mini is 80 percent cheaper than o1-preview, making it a powerful, cost-effective model for applications that require reasoning but not broad world knowledge.

ALSO READ: UIDAI warns users to avoid these common Aadhaar card mistakes

Inputs from IANS

Advertisement

Read all the Breaking News Live on indiatvnews.com and Get Latest English News & Updates from Technology

Advertisement
Advertisement
Advertisement
Advertisement