Sam -German, CEO of Openai and Lisa Su, CEO of advanced advanced devices, witnesses during the hearing of the Trade, Science and Transport Committee in the Senate entitled “Winning the Artificial Intelligence Race: strengthening American capabilities in computing and innovation”, in the construction of Heart on Thursday, May 8, 2025.
Tom Williams CQ-Roll Call, Inc. | Gety pictures
In a sweeping interview Last week, the CEO of Openai Sam Altman addressed a large number of moral and moral questions regarding his company and the famous Chatgpt Ai model.
“See, I don’t sleep well at night,” Altman told the former host of Fox News in an interview with every day. Every day, hundreds of millions of people speak to our model.
“I am really worried not to make mistakes in major moral decisions,” German said, although he admitted “perhaps we will also make mistakes.”
Instead, he said he loses the greatest sleep on “very small decisions” about typical behavior, which could eventually have great repercussions.
These decisions tend to the center of ethics that teach ChatGPT, and what questions make by Chatbot do not answer. Here is a plan for some of these moral and moral dilemmas that seem to keep Altman awake at night.
How does Chatgpt treat suicide?
According to Altman, the most difficult issue that the company is struggling with recently is how Chatgpt is dealing with suicide, in light A family blamed Chatbot for the suicide of their teenager son.
The CEO said that among the thousands of people who commit suicide every week, many of them may have been Talk to ChatGPT at interest.
“You may have talked about (suicide), and we may not have saved their lives,” Altman said. “Perhaps we could say something better. Perhaps we could be more active. Perhaps we could have a little better advice, hey, you need to get this help.”
Last month, Adam Rin a The responsibility of the product and the unlawful death suit Against Openai after their son died at the age of 16. In the lawsuit, the family said, “Chatgpt helped Adam’s activity to explore suicide methods.”
Soon after, in a Blog post Entitled “Helping people when they need them more”, ” Openai Detailed plans to address the shortcomings in Chatgpt when dealing with “sensitive situations”, and said it will continue to improve its technology to protect the most vulnerable people.
How are the ethics of the severity of the ethics?
Another big topic was presented in the sitting interview is the morals and morals that teach and buy ChatgPT.
While Altman described the basic model of Chatgpt as a trainer in the collective experience, knowledge and learning of humanity, he said that Openai must then align some behaviors from Chatbot and determine the questions that he will not answer.
“This is a truly difficult problem. We have a lot of users now, and they come from the very different views of life … but on the whole, I was surprised by the ability of the model to learn and apply an ethical framework.”
When you click on how to determine some typical specifications, Altman said that the company consulted “hundreds of moral philosophers and people who think about technology and systems ethics.”
An example of his exemplary specifications made that Chatgpt will avoid answering questions about how to make biological weapons if users call.
“There are clear examples of the place where society enjoys greatly with the freedom of the user,” Germans said, although it added the company “you will not get everything properly, and also needs the inputs of the world” to help make these decisions.
How privacy is ChatGPT?
Another major discussion topic was the concept of the user’s privacy regarding Chatbots, with Carlson’s argue that the obstetric artificial intelligence can be used for “holistic control”.
In response, Altman said that part of the policy he was pressing in Washington is the “artificial intelligence concession”, which indicates the idea that anything the user says in Chatbot should be completely secret.
“When you talk to a doctor about your health or a lawyer about your legal problems, the government cannot obtain this information, right? … I think we must have the same concept for Amnesty International.”
According to Altman, this will allow users to consult AI Chatbots about their medical history and legal problems, among other things. He added that US officials can currently call the company to the user’s data.
“I think I am optimistic that we are able to make the government understand the importance of this,” he said.
Will ChatGPT be used in military operations?
When asked by Carlson if the army would be used by the army to harm humans, Altman did not provide a direct answer.
“I don’t know how people in the Chatgpt army use today … but I think there are many people in the army talking to Chatgpt to get advice.”
Later, he added that he was not sure “exactly how he felt about it.”
Openai was one of the artificial intelligence companies He received a contract of 200 million dollars From the US Department of Defense to put artificial intelligence to work in the US military. Company He said in a blog post It will provide the United States government up to artificial intelligence models designated for national security security, support and road map information.
How strong Openai?
In his interview, Carlson expected that in its current path, Sam Altman, Sam Altman, can collect on its current path, to what is called “Chatgpt”.
In response, Altman said he used to worry a lot about the focus of power that could result from obstetric artificial intelligence, but now he believes that artificial intelligence would lead to a “huge settlement” for all people.
“What is happening now is that many people use Chatgpt and other chat, and they are all more capable. They are all doing more. They are all able to achieve more, start new works, and reach new knowledge, and this is a good thing.”
However, the CEO said he believed that artificial intelligence will eliminate many functions today, especially in the short term.
https://image.cnbcfm.com/api/v1/image/108143180-1746734828531-gettyimages-2213413650-AI_hearing382_050825.jpeg?v=1757923406&w=1920&h=1080
Source link