According to the CEO of Anthropor Dario Amani, the AI Models (AI) are less than humans. According to the report, the statement was made by the Executive Director of the Opening Company Law with Claude’s event on Thursday. During this event, the San Francisco Intelligence Company has released new models of Claude 4, as well as multiple new possibilities, including improving memory and using tools. According to Amodei suggested that while critics are trying to find road barriers for Amnesty International, “they are not anywhere to be seen.”
The CEO of anthropologist reduces hallucinations Amnesty International
Techcrunch Reports This Amodei made a comment during a press conference, while explaining how hallucinations were not restrictions on artificial intelligence to reach artificial general intelligence (AGI). The answer to a question from the publication, and it was reported that the CEO said: “This really depends on how to measure it, but I think that artificial intelligence models may get less than humans, but they are insulting in more surprising ways.”
Amodei also added that the broadcasters, politicians and human beings participating in other professions make mistakes regularly, so Amnesty International’s mistakes do not take its intelligence, according to the report. However, it was reported that the CEO acknowledged that artificial intelligence models respond with confidence with incorrect responses to a problem.
Earlier this month, Antarubor’s lawyer forced an apology in the courtroom after Chatbot added an incorrect quotation in a file, according to Bloomberg. a report. The incident occurred during the continuous lawsuit of the artificial intelligence company against music publishers due to the violation of the alleged copyright for each of the 500 songs.
In October 2024 paperAmodei claimed that Antarbur may achieve AGI as soon as possible next year. AGI refers to a kind of artificial intelligence technique that can understand and learn knowledge and apply through a wide range of tasks and implement procedures without the need for human intervention.
As part of her vision, Antarubor Claude Obus 4 and Claude Sonit 4 issued a developer conference. These models bring significant improvements in coding, use of tools and writing. Claude Sonnet 4 recorded 72.7 percent on the Swe-Bench index, as it achieved discrimination on the latest Sota in writing code.
