Artificial intelligence seems poised to transform the way humans work and learn, perhaps in less time than many expected.
For now, however, ChatGPT, the artificial intelligence system based on neural models of language, which since it was presented in November by the Open AI company has focused the attention of millions of people around the world for its ability to write reports. or writing poems, he has come across a Selectividad History of Spain exam, and he has not come out very well.
ChatGPT has obtained a 5.3, and has needed a third correction to get the pass.
EL PAÍS has submitted this week the most popular artificial intelligence system to the same Spanish History exam that the students who took the ordinary call for the University Access Assessment (EVAU) took in June.
The newspaper asked two Aragonese History teachers, with experience as correctors in the Selectividad, to try to review the exercises as if they had been done by a student.
The teachers discussed the exam without talking to each other.
The first to deliver it, on Thursday afternoon, was Juan Pedro Serrano.
He gave it a 5.75, and conveyed some disappointment with the bot's performance.
The second, less benevolent correction, was made by Iván Heredia, who rated it 3.75 and began the email in which he attached the revised proof by stating:
“It has been quite an experience correcting this exam.
It's interesting.
I think that if I did not get to know that it is an artificial intelligence, I would have given it a lower grade ”.
The average of both grades gave a fail.
The EVAU regulations provide, however, that when there is a difference of more than two points between the first grade of the exam and the one obtained in the review of the exercise, as was the case, the test must be submitted to a third correction, and the final grade is calculated by taking an average of the three.
Susana Lozano, another Aragonese History teacher who has been a corrector at EVAU, accepted the assignment on Friday without knowing the result of the corrections that her colleagues had made.
She gave it a 6.5, the average went up to 5.3, and ChatGPT ended up just barely passing.
The disparity in the teachers' judgments reflects the confusion generated by the way artificial intelligence responds, with simple and cold sentences, but correct and without the organizational problems and misspellings that second-year students sometimes commit. of high school.
Teachers consider that the ChatGPT exam abuses generalizations, hardly includes examples, has notable shortcomings, is at times excessively "politically correct", and is written in a style that reminds them of a student "who has learned the theme from memory and reproduces it using phrases, words and expressions that you probably don't quite understand”.
The advance of AI, as has already happened with the generalization of the use of Google, also questions the conception of teaching as a mere transmission of content,
Susana Lozano acknowledges that when correcting she had the compassionate point that she would have had towards a student who has not been able to go to class and has taken the exam "on their own".
And Juan Pedro Serrano believes that things will change in a short time: "Probably he has not entered all the contents or all the correction criteria of the EVAU tests that have been in Spain, but I suppose that when he has them he will end up nailing the exam".
The questions were formulated by asking ChatGPT to answer them as if it were responding to a Selectividad History of Spain exam (the process included some stoppages caused by technical problems with the AI, sometimes saturated by the volume of simultaneous requests it receives).
As his first answers were often very short (despite the fact that he had been asked for a larger number of words), he was asked to expand his explanation with elements that he had not already mentioned.
The AI tended to be reluctant, however, and the addition was often largely a repetition of what it had initially written combined with some new bit of information.
That redundancy surprised teachers, and is one of the reasons they find most of your answers weak.
The following is a reproduction of the examination carried out by ChatGPT (the original exercise gave a choice between several options in each section, a selection that was decided by this newspaper), and part of the observations to their answers made by Juan Pedro Serrano (to the questions on the 30 Years War and the Dictatorship of Primo de Rivera) and Iván Heredia (to the others).
Question 1. Short questions to developI) Celts and Iberians on the eve of the Roman conquest.II) Causes of the Thirty Years' War, and its consequences for the Spanish monarchy and Europe. Question 2. Historical Source.
Based on the following text, the student will answer three questions
“I affirm that no political creed, wherever it comes from, even if it was revealed in a burning bush, has the right, in order to gain power, to subject its country to the horrendous martyrdom that Spain is suffering.
The magnitude of the nonsense, the gigantic error, is more easily measured with a less dramatic, almost vulgar consideration.
Two years ago this drama began, apparently motivated in the political order by not wanting to respect the results of universal suffrage in the month of February 1936. Two years have passed.
And it can be argued that […] if instead of committing this madness the normal regime had been followed, at this time it is almost certain that we would be on the eve of a new electoral consultation, in which all Spaniards could freely prove their political forces in Spain.
[…]
It is a profound mystery, in this country of surprises and unexpected reactions, what may result tomorrow when the Spanish, in peace, consider what they have done during the war.
[...] But it is a moral obligation, on all those who suffer from the war, when it ends as we want it to end, to get the greatest good possible from the lesson and the muse of the punishment, and when the torch passes into other hands , to other men, to other generations, who will remember, if they ever feel their angry blood boil and again the Spanish genius rages again with intolerance and hatred and with the appetite for destruction, that they think of the dead and listen to their lesson: that of those men who have fallen raging in battle fighting magnanimously for a grandiose ideal and who now,
Last speech of Manuel Azaña as President of the Republic, in Barcelona, on July 18, 1938, "Peace, Mercy, Forgiveness".
I. Classify the source, and indicate who the author and recipients are II.
Summarize the ideas of the speech and assess the approach it exposes regarding the conflictIII.
It contextualizes the speech and indicates the political evolution of the Republican side during the civil war and the role of President Azaña Question 3. Topic.
The student will develop the following topic: The dictatorship of Primo de Rivera.
You can follow EL PAÍS EDUCATION in
and
, or sign up here to receive
our weekly newsletter
.
Subscribe to continue reading
Read without limits
Keep reading
I'm already a subscriber
