OpenAI introduces new artificial intelligence models after altering their names, the latest AI model series named 01 AI is so advanced it can think before responding. The model is designed to think before answering any question and can answer complex questions.
In a Sep. 12 blog post, the OpenAI stated that the 01 AI model can be accessible today with significant advancements and new levels of AI capabilities. OpenAI explained that these new models are intended to take their time and use ‘chain-of-thought’ reasoning to enable them to solve more intricate tasks, especially in STEM and coding, among others.
To demonstrate what the OpenAI o1 had been designed to do, OpenAI posted two videos of its new model; making a video game out of a simple narration and completing a complex logical as well as mathematical problem. The “preview” and “mini” of OpenAI o1 were released to the ChatGPT Plus users with the promise of better ones in the future.
It has also been noted that o1 seems to perform better than GPT-4, especially in science subjects at PhD level such as Biology, Chemistry, and Physics, and in some of the U. S high school exams.
According to OpenAI, the o1 mini model is trained in STEM-related reasoning, which means that it has more limited information about history biographies or trivia. Its understanding of those non-STEM topics is similar to the other small language models such as GPT-4 mini.
The company was aware of such drawbacks and intends to improve in subsequent versions; the possible future increase in the model’s utility may not be limited to STEM disciplines alone.
Some rumors suggested that OpenAI was planning to release a reasoning-centric AI model known as Strawbery in September, however, OpenAI has not made public, what the differences between the two models it is currently working on are.
