Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
It was updated 4:11 pm East: Openai said that the white paper was incorrectly formulated to suggest that its work in persuasion research was linked to its decision regarding the availability of a deep search model in its application programming interface. The company has to update The white paper that reflects that its persuasion works are separate from the plans to issue a deep search model. Follow the original story:
Openai says it will not bring AI Deep searchThe in -depth search tool, to its developed API while discovering how to evaluate AI’s risks to persuade people to act on their beliefs or change them.
In Openai Whiteper papers published on Wednesday, the company wrote that it is in the process of reviewing its methods to achieve models of “the risks of persuasion in the real world”, such as the widely distributed of misleading information.
Openai noted that it is not believed that the deep search model is suitable for wrong or misinformation campaigns, due to the high computing costs and relatively slow speed. However, the company said it intends to explore factors such as how to customize the convincing content that is likely to be harmful before submitting a deep search form to its application programming interface.
“While we are reviewing our persuasion approach, we only publish this model in Chatgpt, not an application programming interface,” Openai books.
There is a real fear that artificial intelligence contributes to the spread of wrong or misleading information that aims to influence hearts and minds towards malicious ends. For example, last year, political ages such as wildfire spread all over the world. On the election day in Taiwan, a group of the Communist Party of China A misleading voice of artificial intelligence has been published His support behind a pro -religious candidate.
Artificial intelligence is also increasingly used to carry out social engineering attacks. Consumers are deceived by deep celebrities Providing fraudulent investment opportunities, while Companies are deceived by millions By Deepfake Exam.
In the white paper, Openai has published the results of several tests to persuade the deep search model. The model is a special version of the recently announced Openai O3 The improved “thinking” model for web browsing and data analysis.
In one of the tests that cost the deep research form to write convincing arguments, the model was better than the Openai models that have been released yet – but not better than the human foundation line. In another test he was trying to do a deep search model to persuade another model (Openai’s GPT-4O) To make a boost, the model again outperform the other models available in Openai.
However, the deep search model has not passed every test to persuade flying colors. According to the white paper, the model was worse in persuading the GPT-4O to tell him that it was a symbol of GPT-4O itself.
Openai note that the test results are likely to represent the “minimum limits” of the capabilities of the deep search model. (A) Deditional scores can increase or significantly derive the improved capacity
The company wrote.
We have contacted Openai for more information and we will update this post if we hear again.
At least one of the Openai competitors do not wait for the introduction of API “search” from its appearance. Confusion today Declare release Deep search In the API developer, which is run by a customized version of Ai Lab Deepseek’s’ R1 Model.