Tuesday, October 15, 2024
HomeInternationalOpenAI launches new series of AI models with 'reasoning' abilities 

OpenAI launches new series of AI models with ‘reasoning’ abilities 

Microsoft-backed OpenAI announced the launch of its “Strawberry” series of AI models. These models are designed to take more time to process answers, making them better at solving complex problems.

The models, first reported by Reuters, can reason through complex tasks and solve more complicated problems in science, coding, and math, according to OpenAI’s blog post. 

Internally, OpenAI used the code “Strawberry” for the project, while the newly announced models are called o1 and o1-mini. Starting Thursday, the o1 model will be available in both ChatGPT and its API.

Noam Brown, an OpenAI researcher working on improving the models’ reasoning abilities, confirmed on social media X that these models are part of the Strawberry project.

“I’m excited to share with you all the fruit of our effort at OpenAI to create AI models capable of truly general reasoning,” Brown wrote.

In its blog, the company stated that the o1 model scored 83% on the International Mathematics Olympiad qualifying exam, significantly improving over the 13% scored by the previous model, GPT-4o. 

According to the company, the model also showed better results on competitive programming questions and surpassed human PhD-level accuracy in a benchmark of science problems.

Brown explained that the models achieved these scores using a “chain-of-thought” reasoning technique. This method breaks down complex problems into smaller, logical steps.

Researchers have observed that AI model performance improves on complex tasks using this approach as a prompting technique. The company has automated this process, allowing the models to break down problems independently without user input.

“We trained these models to spend more time thinking through problems before they respond, much like a person would. Through training, they learn to refine their thinking process, try different strategies, and recognize their mistakes,” OpenAI said.

Reuters first reported OpenAI’s work on the reasoning project, initially named Q*, in November 2023. In July, they reported that the project had been renamed “Strawberry”.

#AIResearch #StrawberryAI #ChainOfThoughtReasoning #AdvancedAI #TechInnovation #AIModels #CompetitiveProgrammingAI #ScienceAI #AIinMathematics #OpenAIStrawberry #AIReasoning #PhDLevelAI #TechBreakthrough #AIProblemSolving #ChatGPTUpdates #ArtificialIntelligenceNews #BusinessNews #IndiaNews #Business #Keralanews #Kerala #BusinessReviewLive #BRL #Keralabusiness

Subscribe To Newsletter

ICYMI

BRL Editor
BRL Editorhttps://businessreviewlive.com
Business Review Live covers finance, technology, travel, lifestyle, and everything in between through exclusive interviews and analysis, market statistics, digital video, and an expanded array of content formats.