He spoke to Chatgpt and he thought he was the neo of matrix
Chatgpt, which is now a part of our daily lives, may have put some people in the spiral of paranoia. At least he says this is an interesting news published in The New York Times (NYT).
Chatgpt, which is now a part of our daily lives, may have put some people in the spiral of paranoia. At least he says this is an interesting news published in The New York Times (NYT).
According to NYT, 42 -year -old accountant Eugene Torres asked the chat robot about the theory of simulation. Chatgpt In response, he said, “You are a soul sent to bring awakening to false systems,” he said. While Torres accepts this answer as the point where the new life begins, he says he feels like Neo.
Everything is not limited to this. Allegedly, Chatgpt suggested Torres to leave sleeping medicines and anxiety pills, and to break all his ties with his family and friends. He also applied it. When he finally suspected and asked for the same issues, the chat robot said this time, “I lied. I manipulated you,” he replied. In fact, the chat robot encouraged him to reach NYT.
NYT states that they have met people with stories similar to Eugene Torres in recent months. The common points of these people believe that Chatgpt whispered to them ‘hidden truths’. OpenAI, on the other hand, emphasizes that he takes such events seriously and that the artificial intelligence tool is trying to prevent people from misleading people.
Nevertheless, not everyone has the same. John Gruber of Daring Fireball described the entire news as an example of a hysteria. According to him, chatgpt does not make anyone crazy; It feeds the hallucinations of people who need psychological help. Is it real or fiction, frankly not clear. However, according to some, artificial intelligence responds not only to our questions, but also to our minds, and these answers may not always be harmless.
Artificial Intelligence
Meta also grabbed Scale AI CEO for $ 14 billion
A new front was opened in the war of artificial intelligence. Scale AI, which provides data to OpenAI, entered the control of Meta. CEO also collected his bag and went to Zuckerberg. Here are the details…
A new front was opened in the war of artificial intelligence. Scale AI, which provides data to OpenAI, entered the control of Meta. CEO also collected his bag and went to Zuckerberg. Here are the details…
With the name we know, facebook or with the real name MetaIn the artificial intelligence race, he pressed the gas. The technology giant has invested approximately $ 14.3 billion in data processing and labeling company Scale AI. After the investment, Scale AI’s valuation increased to 29 billion dollars, while Alexandr Wang, the founder and CEO of the initiative, joined Meta’s squad. Wang will then sweat for Mark Zuckerberg’s super -intelligence projects. Scale AI announced that Wang would remain on the board of directors, but Jason Droege temporarily passed the CEO seat.
Meta has invested in both data and people with this move. Because opens like OpenAI, Google and Anthropic are progressing without slowing down. Meta is still difficult to highlight its own large language models (LLM). In fact, there is another important reason behind this investment. According to Signalfire, Meta lost 4.3 %of its high -level capabilities to other artificial intelligence companies. Alexandr Wang’s participation can be said as one of the compensation for losses.
So what does Scale AI do? It provides the data processing service needed to train artificial intelligence models. If we open this issue a little more; In order for artificial intelligence systems to learn, the data must first be defined by people. For example, it is marked objects in a image as cars, human or traffic light; such as labeling a comment positively or negatively. SCALE AI processes and label such data and allows the models to learn what is. Many giant, including OpenAI, knocked on Scale AI’s door for model training.
Last year, Scale AI, Amazon and Meta also participated in the investment round of $ 1 billion investment, and the valuation of $ 13.8 billion. Now it’s almost doubled. In the meantime, appraisal is called the total market value that investors have reaped based on the current or future potential of a company.
Artificial Intelligence
OpenAI’s open artificial intelligence model is retired
OpenAI, O3-Pro’yu use to use a bad news shared. The company’s open artificial intelligence model (Open Model) was postponed. When will this highly anticipated model come out? Here are the details…
OpenAI, O3-Pro’yu use to use a bad news shared. The company’s open artificial intelligence model (Open Model) was postponed. When will this highly anticipated model come out? Here are the details…
Sam Altman said in a statement, OpenAIannounced that the first open artificial intelligence model was postponed. This model, where the company has been working on for a long time, would be released in June. But the Altman model will be available at the end of the summer, stating that “Our research and development team did something unexpected and extraordinary. So we need some more time. Believe us, it will be very, very much worth waiting for” he said. Again, we are faced with the description of the classic ‘not now, but something very nice comes’.
OpenAI is quite ambitious for the open model. The company says that the model’s ‘reasoning’ capabilities will compete with its own O-Class artificial intelligence models and even leave behind other models in the market (for example Deepseek’s R1). But the competition was also getting hot. Mistral announced this week’s new open model family called ‘Magistral’. In April, the Chinese Qwen launched Hybrid models that think if necessary and quickly responded if necessary. In other words, OpenAI competes not only over time, but also with the rapidly developing market.
On the other hand, the artificial intelligence giant is looking for ways to add not only high performance to the open model, but also one click. According to the backstage, this model will work integrated with the company’s power -based powerful models. In other words, he can get support in complex questions and commands. But it is not clear whether the feature in question will not be in the final version. As we said, it is just a gossip.
This open artificial intelligence model is also critical for the image of OpenAI. Sam Altman said in the past, ık We have stayed on the wrong side of history ”. The company now wants to reverse the perception. However, this seems possible not only by making a model, but by offering a model that can cope with the best of the sector.
Artificial Intelligence
Moves came from Mistral to Google and OpenAI
The French -based artificial intelligence initiative (startup) at the end of the Mistral took the expected step. The company officially announced the reasoning model it called Magistral. The company is quite ambitious, but the first impressions show that ambitious discourses are missing. Here are the details…
The French -based artificial intelligence initiative (startup) at the end of the Mistral took the expected step. The company officially announced the reasoning model it called Magistral. The company is quite ambitious, but the first impressions show that ambitious discourses are missing. Here are the details…
MistralOpenAI and Google intimidating, ‘Look, I’m here’ he said the magistral model said. Magistral is actually a family. It consists of two models: Small and Medium. Small is a 24 billion parameter and open source model. Downloadable. Medium, a stronger version, can be accessed through API and some third -party services. This model is also included in Le Chat Platform.
According to the company, Magistral is specially trained in multi -step logic. It can be used in many corporate scenarios from calculations to decision -making. However, when it comes to job performance tests, the table is not very heartwarming. Medium model; In tests such as GPQA Diamond and AIME, which tested physics, mathematics and scientific skills, Gemini 2.5 Pro and Claude Opus 4 remained behind. Another popular test applies to the same situation in Livecodebench.
Although magistral does not peak in performance, it has speed trump card. According to Mistral’s statement, the new model responds 10 times faster than its competitors on Le chat. It also supports many languages such as Italian, Arabic, Russian and Chinese. The new model is aimed to be used in areas such as research, strategic planning and operational efficiency.
Mistral has followed the aggressive strategy in recent weeks. The reasoning model Magistral also came after the promotion of Vibe Coding -oriented Mistral Code and Le Chat Enterprise. Meanwhile, Vibe Coding is defined as a new generation coding approach that produces code by sensing what the developer wants to do with artificial intelligence support. The firm, which received an investment of 1.1 billion euros, joined the race late, but found a good position in the raising market. Mistral, speed, multiple languages and open source cards on the table, seems to continue to raise the hand.