Operai presents AI technology for 'natural conversation'

When Operai began giving private demonstrations of his new GPT-4 technology at the end of 2022, his skills even shocked the most experienced ai researchers. You could answer questions, write poetry and generate computer code so that it seemed far ahead of your time.

More than two years later, Operai has launched his successor: GPT-4.5. The new technology means the end of an era. Operai said that GPT-4.5 would be the latest version of his chatbot system that did not make “reasoning of the thought chain.”

After this launch, Openai technology can, as a human, pass a significant amount of time thinking about a question before answering, instead of providing an instantaneous answer.

It is unlikely that GPT-4.5, which can be used to feed the most expensive version of Chatgpt, generate so much emotion in GPT-4, largely because ai's research has changed in new directions. Even so, the company said the technology “would feel more natural” than its previous Chatbot technologies.

“What distinguishes the model is its ability to participate in warm, intuitive and natural conversations, and we believe it has a stronger understanding than users mean when they ask for something,” said Mia Glaes, OpenAi's research vice president.

In autumn, the company introduced the technology called Openai O1, which was designed to reason through tasks involving mathematics, coding and science. The new technology was part of a broader effort to build which can reason through complex tasks. Companies such as Google, Meta and Deepseek, a new Chinese company, are developing similar technologies.

The objective is to build systems that can carefully and logically solve a problem through a series of discrete steps, each of which is based on the latter, similar to how humans reason. These technologies could be particularly useful for computer programmers who use ai systems to write code.

These reasoning systems are based on technologies such as GPT-4.5, which are called large language models, or LLMS

LLMS learns your skills by analyzing huge amounts of text extracted from the Internet, including Wikipedia articles, books and chat records. By identifying patterns in all that text, they learned to generate text on their own.

To build reasoning systems, companies placed LLM through an additional process called reinforcement learning. Through this process, which can be extended for weeks or months, a system can learn behavior through extensive tests and errors.

When working through several mathematical problems, for example, you can learn which methods lead to the correct answer and which are not. If you repeat this process with a lot of problems, you can identify patterns.

Operai and others believe that this is the future of the development of ai. But somehow, they have been forced in this direction because they have run out of the Internet data necessary to train systems such as GPT-4.5.

Some reasoning systems exceed ordinary LLMs in certain standardized tests. But standardized tests are not always a good judge of how technologies will work in real world situations.

Experts point out that the new reasoning system cannot necessarily reason as a human. And like other chatbot technologies, they can still be wrong and do things, a phenomenon called hallucination.

Openai said that, as of Thursday, GPT-4.5 would be available to anyone who has been subscribed to Chatgpt Pro, a $ 200 service a month that provides access to all the company's latest tools.

(The New York Times sued Openai and his partner, Microsoft, in December for the infringement of copyright of the news content related to ai systems).