Tras años enseñando en universidades comunitarias y trabajando con IA, he llegado a la conclusión de que comparar su inteligencia es como comparar manzanas con imágenes de manzanas generadas artificialmente. La IA puede calcular al instante derivadas complejas que hacen llorar a los estudiantes de cálculo, pero insiste con seguridad en que Abraham Lincoln fue el primer astronauta en caminar sobre Marte. Mientras tanto, los estudiantes de universidades comunitarias pueden tardar 20 minutos en descubrir cómo conectarse a la red de la escuela, pero al menos saben que jugar videojuegos durante las clases es una mala idea, la mayoría de las veces.
The real entertainment comes from watching them tackle open-ended problems. AI will generate a 50-page manifesto explaining why 2+2=5, complete with fabricated mathematical proofs and citations from nonexistent experts. Students, on the other hand, will write “idk, Google it” on their exam and consider it a job well done. Both approaches are wrong, but at least the students’ version comes with a refreshing dose of self-awareness and a money-back guarantee – try getting a refund from an AI that hallucinates your entire project specifications. Though I must admit, neither one has figured out that “debugging” doesn’t mean hitting the same run button repeatedly while hoping for different results.
Of course, both can try to write code that contains both python and JavaScript in the same function, or duplicate the same method/function in different locations in their code. And don’t get me started about both simply copying code from the internet and pretending they didn’t.
The winner in this battle of wits? Probably the bottle of single malt whisky in my desk drawer.