A report shows that AI is still below human knowledge

a report shows that ai is still below human knowledge.jpg
a report shows that ai is still below human knowledge.jpg

AI human knowledge

Is AI ready to surpass human knowledge? ChatGPT is one of the most used virtual tools in recent months to ask any type of query or ask them to perform a task for us in a matter of seconds. However, research has questioned its effectiveness by detecting that It did not execute some functions correctly that the human being is capable of doing. What exactly is it about?

The Artificial intelligence has represented a true technological revolution, far surpassing the way in which we communicate through its ChatGPT consultation system, especially in recent months, although many aspects of its functional structure still need to be polished, since experts verify that It is in an early phase that does not fully exploit its capabilities.

Therefore, a investigation carried out by the Rovira i Virgili University (URV), accompanied by the Humboldt University of Berlin, the Autonomous University of Barcelona (UAB) and the Catalan Institution for Research and Advanced Studies (ICREA) has taken the approval of the language used by ChatGPT in grammar corrections. What’s the score?

The AI ​​does not pass in grammar

This study has been carried out with the objective of immediately identifying errors in different types of sentences to see if they are well formulated in their native language or not. For this, the systems have been used ChatGPT, GPT3 and GPT3.5. In this way, the test between humans and machines consisted of asking a simple question: “Is this sentence grammatically correct?”



AI human knowledge

The experts concluded that human intelligence is still above artificial intelligence, since they obtained many wrong answers from the large language models, even adopting a monosyllable “yes” achievement, whether the answer was correct or not, without arguing his reasoning. The leader of the study, Vittoria Dentella explained that the result was surprising and believes that there is a double mismatch between humans and AI.

Thus, it was ruled that people do not have access to negative evidence, to know that a phrase is poorly written or said poorly in any language, while machines are based on human feedback to access this information, making it difficult for them to recognize common grammatical errors.

Dentella concludes that using this type of tools to promote linguistic abilities similar to those of humans is not the most advisable, since this language model It is not optimized enough in its current stage of development.

Productive uses in ChatGPT

However, ChatGPT has managed to exceed its expectations since it has been created to perform a wide variety of tasks that make life easier for the user and promote considerable time savings. It is true that anyone can use this system using natural language, although it must be considered that specific and well-written questions must be asked to get accurate answers.

Although Artificial Intelligence has surprised many users and experts after more than a year operating on computers around the world, the truth is that ChatGPT, owned by Open AI, is a sophisticated tool with a lot of material to improve. and several years of development ahead.

Previous articleYou will be able to chat with Messenger without a Facebook account, at least in Europe
Next articleSamsung is closer than ever to Android: the Galaxy S24 is almost a Google Pixel
Expert tech and gaming writer, blending computer science expertise