Metacognition, the great frontier of AI/Metacognición, la gran frontera de la IA

in #piotr14 days ago


As much as they want to scare us with artificial general intelligence (AGI), we all know that this is not going to be achieved with these language models like chatGPT or Gemini, which at the end of the day are nothing more than sophisticated search engines that, Instead of giving you 354,453,214 links for you to search, it summarizes it for you in half a dozen paragraphs.

Por mucho que nos quieran asustar con la inteligencia artificial general (AGI), todos sabemos que eso no se va a lograr con estos modelos de lenguaje del estilo de chatGPT o Gemini, que al fin y al cabo no son más que buscadores sofisticados que, en lugar de darte 354.453.214 enlaces para que te busque la vida, te lo resume en media docena de párrafos.

And it's not because I say so, more and more AI experts consider that continuing to increase large language models has a limit beyond which no further progress will be made, some even think that their expiration date will arrive before the next five years and therefore different approaches are beginning to be considered.

Y no es porque lo diga yo, cada vez son más los expertos en IA que consideran que seguir aumentando los grandes modelos de lenguaje tiene un límite a partir del cual no se avanzará más, incluso algunos piensan que su fecha de caducidad llegará antes de los próximos cinco años y por ello comienzan a plantearse distintos enfoques.


Among these experts is Bill Gates, who is more dangerous than a piranha in a bidet, but it must be recognized that he is a brilliant mind, who advocates what they have called "Metacognition" or, in other words, the ability of an artificial intelligence system to reflect on its own thought process.

Entre estos expertos se encuentra Bill Gates, que tiene más peligro que una piraña en un bidet pero hay que reconocer que es una mente brillante, que aboga por lo que han dado en llamar la "Metacognición" o, lo que es lo mismo, la capacidad de un sistema de inteligencia artificial para reflexionar sobre su propio proceso de pensamiento.

To achieve this, AI should know how well it is performing a task and detect when it makes mistakes, understand its own limitations and be aware of what it knows and doesn't know or when it needs more information or assistance and, of course, know how to explain how it got there to a certain conclusion and justify their decisions.

Para lograr esto la IA debería saber qué tan bien está realizando una tarea y detectar cuándo comete errores, entender sus propias limitaciones y ser consciente de lo que sabe y no sabe o cuándo necesita más información o asistencia y, por supuesto, saber explicar cómo llegó a una determinada conclusión y justificar sus decisiones.


Being aware of its limitations would help it avoid making mistakes and provide more reliable information and by learning from their experiences it can adapt to new situations and tasks. On the other hand, by being able to explain its reasoning it could gain the trust of humans and collaborate with them more effectively.

Ser consciente de sus limitaciones le ayudaría a evitar cometer errores y proporcionar información más confiable y al aprender de sus experiencias puede adaptarse a nuevas situaciones y tareas, por otra parte, al poder explicar su razonamiento podría ganarse la confianza de los humanos y colaborar con ellos de manera más efectiva.

It seems that there is no doubt that metacognition has the potential to revolutionize the way we interact with AI, increasingly artificial intelligence will function more like us humans which gives me a rather bittersweet feeling, like when your girlfriend tells you: "Don't worry, of all your friends you are the one who makes love the best."

Parece que no cabe duda de que la metacognición tiene el potencial de revolucionar la forma en que interactuamos con la IA, cada vez la inteligencia artificial funcionará más como nosotros los humanos lo cual me produce una sensación bastante agridulce, como cuando tu novia te dice: "tranquilo, de todos tus amigos tu eres el que mejor hace el amor".

More information/Más información


Honestly brother AI may own it all in our future hahahaha. I find the idea of AI reflecting on its own thinking pretty interesting. Imagine AI recognizing its errors and learning like us humans. There may be a point where you would not have to Work on advancing technology, it would probably advance itself

Congratulations @mauromar! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)

You have been a buzzy bee and published a post every day of the week.

You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word STOP

Check out our last posts:

Our Hive Power Delegations to the June PUM Winners
Feedback from the July Hive Power Up Day

We shall see as things progress. An honest to God program that is able to learn from mistakes would be a huge jump, however, it won't make it self aware. Just another more advance search engine!

We definitely need to be aware of these types of Progression at times because it can also be quite negative