Introduction

My name is Carlos Girón, I'm part of the SAPiN team and my background is from philosophy actually so yeah it's kind of strange to be here in an AI event but well I think we will see it's not so strange at the end.

Reframing “the Future of AI”

So what I want to talk about is the future of AI and yes the future of AI is a question so we can understand this phrase in at least two ways. The first one, it's pretty simple, we don't know what is going to happen in the future So it is an open question, right?

But that's not what I want to say today to you What I want to leave you is the idea that the future is actually the question

questions, it's actually what it's going to be in the future of AI and this supposed to have kind of a change in our mindset because

AI Is Here—So It Has a Future

Some years ago And this goes very fast as we already saw ago, we used to think about AI as something that is in the future.

And suddenly, AI is here. It's no longer in the horizon. It's here in our hands, literally.

We have it in our hands. And the thing is that now AI has a future. It's something that has a future.

So this changes the question by itself.

Responsibility Shifts to Users

Why? Because if this is something that has a future, with a future, then we have to ask, who's responsible of it? Who's responsible of the future of these things that we have now in our hand and the answer of course is in one part the owners of the technology they have a role in this but what is more important for me is that we must be we must be conscious that we as users have a very important role in this and we as users we do questions so that's what the issue is here why there are questions create better intelligence

So if we focus back responsibility, if we recover responsibility to us as human beings, two things will happen. First, our approach within AI will change completely because it's not the same thing to have something that you know you are responsible about. That's something that is there in the horizon.

Right. So for those of you or maybe in the audience that think that I thought sometime that the future of AI was something like this. Sorry to disappoint you, it is not like that. No, this is good science fiction, but for now, we are still safe, right?

Why? Because the future of AI, this thing that has a future now, and it's in our hands, has no will, right? So we are the ones that have a will. We have something, some ideas, and we push the AI in a way, right?

The Power of Questions

So there's a strange pattern, and I don't know if you have noticed of course yes you have noticed this but for

Vague, Biased, and Thoughtful Questions

example if I do a big question to the models to the language models for example I get a big answer so things like tell me about AI how can I improve my life yes people is using AI for this of course give me some ideas for a project some ideas just like that what should I do in this situation explain

marketing but what you can see here and if you can notice this pattern is that that we are used to ask like this to browsers. We are used to ask like this to, I go to YouTube, I go to whatever thing has a content and I can search for, and I do this kind of questions.

So they are valid questions, yeah? They are vague questions, but valid questions. And we are used to work here, and we are shifting to something else very quickly.

We have biased questions. This is more complicated, and we will speak about this in an event next week.

but if I say something like why is remote work clearly better than office work if I phrase the question like this they are all now we have clearly better

done I have already a bias right I do you think AI will replace all human jobs there's already a bias because I am actually assuming this is going to happen right and if I write like this a question to the large language models

models, what is going to happen is that the model will understand my point, and it will answer from this place, right?

So it's no surprise that if I ask something like this, I will have an answer like, well, yes, most probably, it will happen.

And then I will say, oh, yes, those Elon Musk and those guys are the worst in the world. Thoughtful questions,

then we have to move forward this and again don't get me wrong with this all of these are valid questions the things I have to be conscious about what I am asking and from which position I am asking but if I change to this what are

the main trade -offs between remote and in -office work under what condition does AI replace jobs and so on so forward you have if I have these thoughtful questions now then suddenly it will appear up apparently will be a dialogue with an intelligence farther away.

But it didn't change. The machine is the same. What changed is the question. So this is the main idea for today.

Why Prompt Quality Determines Outcome Quality

1The impact of AI will depend less on machine intelligence and more on the quality of human questions. This is something we can work with, right?

And it's very important, Cesc already said it. We saw the process, we saw that thing, but the main thing and the main goal and the main asset at the end is the first point,

how you did the question, how you did the prompt. And that is a very high value thing.

I read this phrase and I really liked it in Nico Roth's, the quality of your results is tied to the quality of your questions. But this was from a book that was related

to Google Analytics. So since then, we already knew this.

we already knew that data is going to be there we have data everywhere we can extract data from everywhere right now that's not the issue the issue is if I have all this data and I don't have the right questions then it's useless right

and that's the issue we still need to improve our ability to make better questions and therefore to think AI as a thinking interface to think about this as this.

AI as a New Kind of Tool: A Language Interface

This is another point that is important and there's the concept of tools here.

I'm going to go faster in this point because philosophically speaking this is a discussion. What is a tool and what not?

Just to understand the

old way and the new way we can think about the hammer, right? If I grab a hammer and I just hit the nail and get it in the well, this is more or less how how it works with computers, more complex of course, until now, yes?

What is changing is now I'm not clicking here, I'm not doing anything, an input, the only input I am doing with AI now is I'm speaking, so it's magic because it's like if I was able to say, hammer, please put the nail in the wall and just leave a couple of centimeters so I can hang my photograph, please. And without doing anything, it happens.

So the big change here is that our relationship, our interactions are with language, with natural language. And that's a big, big, big change in how we are approaching knowledge in general.

From Asking Questions to Giving Instructions

So the skill of the AI age will, tools will change quickly, we already saw it. New platforms will emerge, but one fundamental skill will remain essentially thinking clearly enough to

give good instructions and now we are shifting from questions to instructions and that's different to do make a question is different than delivering an instruction with an instruction I know what I want with a question I don't know exactly what I want but also when we use language which is the shift to AI as a a thinking interface, something else changes.

Prompting as Clarifying Human Intention

First of all, we're importing all the good stuff about language, but also the bad stuff about language. And the first thing that is not so good about language is that we don't understand always what we want to say. We are very clear always, right?

If you have a couple, you know this perfectly. A way of clarifying what we actually want. this is prompting this actually a good prompt is a way to clarify what you want and say it clearly it sounds simple but we all know sometimes it's not as simple as it sounds so a prompt is simply human intention translated into clear

instructions right and this Socrates knew about this back to Socrates we can find something he used to show to teach people how to ask better questions and So artificial intelligence forces us to do exactly the same. That's why I think about prompting as philosophy 5 .0.

So language shapes our thought. Ludwig Wittgenstein used to say the limits of my language mean the limits of my world. And to explain this, I will just share a little thing, a little history.

If you see this picture, you will see the moon, right? When I arrived to Barcelona, because I'm from Mexico, on one of those super moons days, beautiful moons I said oh yeah look at the rabbit in the moon and my friend said like what the fuck is this right what is this what's the rabbit in the

moon well there's the rabbit in the moon and this is because there's a legend in Mexico about a God taking the rabbit to the moon so the concept of rabbit in the moon is in my head and it determines what I see in the moon right so this happens constantly in our lives.

Mindset Matters: Cognitive Attitudes

So the problem of today is that this kind of stuff, that when this happens, what is happening here is that we have a cognitive attitude. That's what I call a cognitive attitude, which is the mental posture from which we approach a task, right?

The invisible mindset shapes the language we use and ultimately determines the quality of the result we receive.

So if I am conscious that the rabbit in the moon is not something for for everyone, then I will communicate better. I will be more conscious about it, and so I will communicate better.

Ego, Curiosity, and Methodical Doubt

So I want to give you three cognitive attitudes, ego, curiosity, and methodical doubt.

So when I approach something, when I approach a task in AI from here, I can say something like, give me exactly this.

It is not bad. It is good for something. It depends on what's your objective.

From curiosity, the same thing, propose different approaches for this thing.

And for methodical doubt, ask questions before answering and then I will get an idea, right?

So we have three different things and the important thing in the future of AI is not to avoid this and go there. This is not the issue.

The issue is to develop the ability, the plasticity, the cognitive plasticity to go from one to another depending on what I want to achieve and this is something we have to train as

humans so we can think about them as characters not as the ultimate truth

Practical Example: Planning a Weekend in Barcelona

right so one last practical example if we plan a weekend in Barcelona from ego approach I will say give me two day itinerary for Barcelona with five attractions the result will be something like go to Sagrada Familia, Paroel, called the Ramblas.

This is good, right? It's not bad, but it could be better.

Curiosity approach. Curiosity approach, propose three different itineraries for Barcelona, cultural healing gems, and relax it, and the result will be market, local, neighborhood. It could be surprising, this one, right?

And then the methodical approach, before proposing an itinerary, ask me five questions about my interest, budget, and travel style, and then I will have a tailored plan so it depends on what you want and sometimes in a project it will depend on which stage you are that one approach will be better than the other one right

Conclusion: AI as a Telescope for the Mind

so to close this artificial intelligence does not just generate answer it amplifies the structure of human thinking and just already said it if if your thinking is a mess it will be amplified also so you have to think clearly because if you have a mess in your head it will be amplified as well

Because AI has no will, again. The clarity of your answers is a reflection of the clarity of your thinking. 1Artificial intelligence is like a telescope for the mind. That's the main issue for me today.

So it's a powerful instrument, yes, but it does not decide where to look. It only amplifies the direction we choose. So this will be a better image of what the future looks like.

this poor AI taking quick notes of what we are teaching to the AI yeah not that scary robot in at the beginning so AI is a telescope for the mind what we discover depends on where we point it I hope you find this useful and thank you very much so you have questions