Skip to content

Newly released AI surpasses 90% of humans

Newly released AI surpasses 90% of humans

The latest artificial intelligence model follows a multibillion-dollar partnership between OpenAI and Microsoft

A new artificial intelligence program can keep up with humans in a series of professional and academic tests, according to its creator, who said the model scored in the top 10% of participants in a mock bar exam.

Technology firm OpenAI unveiled its latest language model, called GPT-4, on Tuesday, saying the program is “more reliable, creative and capable of handling much more nuanced instructions” than its predecessor, GPT-3.5.

“GPT-4 is a large multimodal model (accepts image and text input, emits text output) that, while less capable than humans in many real-world scenarios, shows human-level performance in several professional and academic benchmarks,” he said. “For example, he passes a mock bar exam with a score in the top 10% of participants; in contrast, the GPT-3.5 score was about 10% lower.”

The AI ​​model also performed in the 93rd percentile on an SAT reading exam and in the 89th percentile on an SAT math test, the company added.

Read more

Will humans become an extension of their machines? ChatGPT can have irrevocable consequences for learning and decision-making

The GPT software has been integrated into several other applications, including the language learning program Duolingo, which aims to create conversational bots, as well as automated tutors for online education company Khan Academy .

The previous iteration of the model, GPT-3.5, gained popularity in the form of the chatbot program ChatGPT, which is capable of having complex, human-like conversations with users.

According to OpenAI, GPT-4 is its “most advanced system to date” and, unlike GPT-3.5, is capable of processing image requests in addition to text. However, despite its improved capabilities, the company warned that the new model is “not entirely reliable” and still suffers from some bugs, including what it calls “hallucinations,” in which the AI ​​simply fabricates information or generates wrong answers.

READ MORE: AI ‘stresses me out’ – Musk

The company partnered with Microsoft earlier this year, receiving a multimillion-dollar investment from the tech giant to further develop its AI models. The new GPT-4 program will play an important role in Microsoft’s Bing chatbot, which was introduced in a limited way earlier this year, and the company is expected to announce integration into other consumer products in the coming weeks, according to the Financial Times. .

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *