I sense that these days (December 2022) you will be saturated with reading and watching content about ChatGPT and other types of artificial intelligences. Therefore, in an exercise of empathy, I will omit any introduction about it and go straight to the point.
For this reflection we will consider the scenario where the computational capacity or operational costs of AI are as viable as those of the ubiquitous search engine.
In my view, GPT-3 and other types of language models will have a transformative and unprecedented role as assistants to people but they can never replace, at least completely, the role that Google plays today in society and our lives, in the same way and for the same reason that I think Google will never achieve the utopia of the single result.
Apparently, both aspirations are incompatible with the essential dynamics that define us as humans and social beings.
Let's see.
Both cases are intended to generate a single, correct answer to a question or request for information.
Up to now, when a query is made, Google returns a selection of results sorted by relevance. To order them, it uses certain evaluation criteria, some of which are its own, such as usability or the semantic breadth proposed by the content, and others are social, which it extracts from the Internet community itself, such as links from other pages or the behavior of users when consuming it.
Already in itself, this system can be more delicate than we might think, since it limits and manipulates our access to information by choosing under its criteria what is relevant and what is not.
This generates situations where the system could be abused or cases where the results of a query do not include that brilliant article written in an unknown blog or with insufficient authority to rank. It also restricts our ability to access more creative and less validated or normative information.
On the other hand, even when typing the same query, users' search intentions vary. In many cases with small nuances. This is why Google shows different types of content in its results.
Whatever the criteria for evaluating information, they can never be to everyone's liking.
But I think the key issue in all this is that information is not the same as knowledge.
As we saw, establishing a relevance for information already implies knowledge. But giving a single true and unique answer would be knowledge itself.
And here the problem begins to become evident, because the truth is that, as far as the human being is concerned, there is no absolute truth. Hopefully, there is general consensus on certain things.
But I find more reasons that lead me to the conclusion that the unique result is not possible without other alternatives.
1. AIs that drink from human knowledge itself can only expose that which has been generated and previously validated by human beings in order to offer answers.
In order to give a unique answer, they will be forced to be permanently contained in that area of knowledge with relative general consensus and never reach the heights of creative knowledge.
The relativization of certain topics would be uninspiring, so much so that people would look for other sources of consultation that offer more interesting and attractive alternatives.
Even the most consensual issues are not absolute.
Neither the universal laws of physics, nor the arrival of man on the moon, nor the appropriateness of a given Law of Law are absolute questions.
Therefore, we do not draw unique conclusions from our own area of knowledge either. Everything is under continuous revision.
No one wants a doxographical exposition with all the interpretations, meanings and approaches to a thing.
It is not possible not to get involved, nor to eliminate any value judgment in the synthesis of the information, because to do so would mean quoting all sources and versions. This is something that nobody wants to consume. In the same way that no one wants or intends to read and delve into all Google results for a search.
On the other hand, generating an argumentative or conversational response and citing only certain sources or authors would directly attack our beloved biases and prejudices.
4. The knowledge of the human being does not have as its sole raison d'être the attainment of an objective and complete truth.
But rather meaningful, responding to more complex human needs and social dynamics beyond describing a strictly natural reality of things. Belief in God could be an example of this. Does God exist?
I asked ChatGPT. His summary answer:
The existence of God cannot be conclusively proven or demonstrated, and each person must come to his or her own conclusion on this question based on his or her own reason, experience and personal beliefs.
Are you satisfied?
If, on the other hand, he had said "Yes." or "No." Would you believe him?
5. The human being will find useless any knowledge that belongs to or emanates from an area external to human knowledge.
All knowledge that does not emanate from the human being will be artificial knowledge, meaningless, unknowable, and incomprehensible to man.
For practical reasons, I prefer not to go deeper into this idea, but I find it a vibrant reflection, full of meaning, and I invite you to evaluate it calmly.
6. Mistrust in the face of abuse.
If we were in a situation without alternatives where we only have a Chat-GPT to make our queries online, we would feel a deep distrust of the possibility of abuse or instrumentalization of the tool for self-interested or perverse purposes.
BONUSThere is a risk of playing the card game.
Provided that an AI was able to educate itself about the individual with whom it interacts, it could give a unique personalized and convenient result for that person, i.e., tell us exactly what we want to hear. A different answer for each person to the same query.
This would circumvent the problem of absolute truth. But honestly, I don't think that's feasible either, because hey, what can I say, we're not AIs, but we're not that dumb either.
CONCLUSION
With this technology in the form of superlative assistants, we will begin to solve much more complex questions and tasks in a matter of seconds. Not only that, AI will inspire designers, filmmakers, lawyers, doctors, researchers, educators, and other professionals. It will help us understand things that we couldn't understand so easily before, and it will bring us closer to human knowledge than ever before. And I definitely think it will be transversally transformative for today's society.
But with this article, dear reader, I invite you not to feel too uneasy. Because in our particular and wonderful existence it seems that only a human being can be. And certain mechanisms of our essence cannot be ignored so easily.
