Artificial intelligence Claude successfully passed the exam in law and economics

News » Science & Technology

< /p>

Economics professor Alex Tabarrok of George Mason University said Anthropic's “Claude” artificial intelligence model successfully passed the exam in law and economics. The examiners noted the low level of knowledge of AI, but it turned out to be sufficient to obtain the maximum passing score. In particular, teachers paid attention to the ability to defend one's point of view with arguments. At the same time, the researchers say, Claude is head and shoulders above the popular ChatGPT – although inferior to him in other respects.

The artificial intelligence, which received funding from FTX founder Sam Benkman-Fried, who is under investigation for alleged fraud, was taking an exam in university-level law and economics, according to Professor Alex Tabarroca of George Mason University in Virginia. At the same time, some of his answers were “better than many human ones.”

Tabarrok noted that there were some flaws in the answer, including that it was an “opinion.” Ideally, the answer should contain more “economic arguments”.

Claude is developed by Anthropic, an AI security research company. In May 2022, the company raised $580 million in funding from investors. Unlike some competitors, this AI does not operate in terms of “good” or “bad”; and “forbidden”, it simply uses the entire set of available data.

AIs avoid talking about controversial topics, which makes them convenient for widespread use, but not needed during the solution of critical and controversial problems. Claude, on the other hand, is willing to give his opinion on almost any topic, but at the same time, he can easily refuse to chat if he considers the request meaningless or provocative.

Claude is more like an artificial person, while ChatGPT is a typical intelligent tool.

Recall that OpenAI, which has already made a splash with its ChatGPT, is training a new neural network to replace entry-level programmers.

Follow us on Telegram