Opeli, a company standing behind the popular AI Chatbot Chatgpt, introduced its own The latest model of artificial intelligence, OpenAI O1Designed for leading in complex tasks of reasoning in science, coding and mathematics. This new model is aimed at solving some restrictions visible in previous models by improving the way AI “thoughts” before generating the answer.
In the Opeli statement, he explained that the O1 model was trained to spend more time considering problems, imitating people's reasoning. “We trained these models to spend more time thinking through problems before they answer, just like a person,” said the company. This approach, which Openai calls “reasoning”, allows the model to improve its thought process, consider many strategies and recognize errors before presenting the answer. The goal is to increase accuracy, especially in more complex tasks.
The new O1 series is the OpenAI response to the constant challenge of AI models generating incorrect or misleading information, often called “hallucinations”. It was a lasting problem with previous models, such as GPT-4, which, although powerful, struggled with complex tasks of reasoning and sometimes gave incorrect answers with great certainty.
According to OpenAI, O1 has already shown impressive possibilities in internal tests. It works at a level comparable to doctoral students on difficult problems in physics, chemistry and biology. In a qualifying exam to the prestigious International Mathematics Olympics, O1 correctly solved 83% of problems, which is a significant improvement compared to 13% of GPT-4 success.
The potential applications of the model are huge. Opeli emphasized its use in healthcare for the annotation of data sequencing data and in physics to generate complex mathematical formulas used in quantum optics. This improved reasoning can bring the benefits of a field that requires multi -stage work flows or detailed analytical work, such as scientific research, software development and engineering.
Opeli emphasized the importance of security when it introduces a new generation of AI models. The company claims that the O1 model is more solid when it comes to compliance with security guidelines, especially in scenarios in which users are trying to bypass their security measures, called “jailbreaking”. In tests, the O1 model exceeded the GPT-4, based on Jailbreak attempts, winning 84 in a difficult safety test compared to the result 22 of the GPT-4.
The company also concluded contracts with the AI Safety Institutes in the USA and Great Britain, granting them early access to the assessment and testing model. Opeli cooperates with both governments to ensure the development and implementation of AI technology in accordance with strict safety standards, especially considering the potential AI risk in areas such as disinformation and bias.
Despite these progress, the growing demand for energy created by large AI models, such as OpenAI O1, becomes a serious problem. These models require huge amounts of computing power to generate complex answers, which in turn increases energy consumption.
Openai progress paves a way that companies such as QDATA are aimed at improving everyday user experiences. Thanks to the QDATA AI assistant, available for free, users can engage in personalized and interactive chats with non -standard chatbot based on chatgPT. Regardless of whether you are looking for information on a specific topic or just want to have a free conversation, Qudata Ai Chatbot can provide intelligent, conscious contextual answers on a wide range of topics.
The OPENAI O1 edition is just the beginning of what the company defines as a new era of artificial intelligence of reasoning. This is the first model in the O1 series, and OpenAI expects regular updates and improvements. Although the current model lacks certain functions that made chatgpt so popular – such as browsing the transfer of internet files and handling files – it is expected to evolve quickly.
