The future is not written, will an algorithm write it?

LA TERCERA – English translation of an opinion article published by ABC on April 14th, 2023

«”Writing is power,” said Margaret Atwood, and Umberto Eco reminds us that, “to survive, you have to tell stories”. Writing helped Anne Frank “let go of everything, her sorrows disappear and her courage reborn.” We write because we feel (“words are written tears,” Coelho said); we write because we think; we write because we live (“to taste life twice, in the moment and in retrospect,” said Anais Nin); We write because we are »

It seems that Homo Sapiens have existed for about 300,000 years. Although we invented the written language about 6,000 years ago, it took more than 5,000 years until its democratization with the invention of the printing press by Gutenberg in the fifteenth century. The profound impact of the printing press on the dissemination of ideas, access to knowledge and the advancement of science is unquestionable.

More than 500 years later, in 1989, Sir Tim Berners-Lee invented the World Wide Web while working at CERN. It marked the beginning of the so-called digital revolution that has universalized access to information and knowledge in digital format to anyone on the planet. It has also allowed the generation and sharing of content to move from being in the hands of a minority to being possible for anyone. It is not surprising then that today about 2.5 quintillion (10^18 zeros) bytes are generated every day, the equivalent of 67 million 90’ movies.

Three decades before the invention of the web, the term “Artificial Intelligence” or AI was born at the 1956 Dartmouth Conference, organized among others by Marvin Minsky, from whom I learned during my time at MIT. The origin of AI is even earlier, dating back to the 40s and 50s of the twentieth century, when research began on the creation of systems that perform tasks typical of human intelligence, such as perception, reasoning and learning. The British mathematician Alan Turing is considered the father of Artificial Intelligence. In 1950 he proposed the famous “Turing test” to determine if we had reached Artificial Intelligence. In this test, a human judge interacts with a machine and a human through text messages in the form of chat. If the judge is unable to distinguish when he interacts with the human and when he interacts with the machine, then the machine is said to have passed the Turing Test and is therefore intelligent. For decades, the Turing Test has been considered the standard for evaluating AI systems. Until recently, passing the Turing test was part of science fiction…

The advances of the last 15 years in AI techniques known as deep neural networks or deep learning, and especially the recent progress in generative AI models are revolutionizing society, similar to how the printing press or the web did. When these generative AI models create text, they are known as Large Language Models (LLMs). They consist of immense deep neural networks trained with billions of examples of texts, mostly from the Internet, including digitized books, newspapers, magazines, and encyclopedias. These models can learn the patterns that govern language, any language, with a level of proficiency similar to or superior to that of humans. In fact, they are proficient in dozens of human languages, as well as programming languages or any other language that exists in nature, such as DNA.

For the first time in our 300,000-year history, we live in a time when non-human entities (computer programs) can create textual content (and also images, music and videos) indistinguishable from the content that a human would generate, but with a scale and speed far superior to those of humans. The potential is virtually limitless.

In a matter of weeks, hundreds of millions of people have begun to interact with chatGPT (the most popular chatbot based on an LLM called GPT-3 or GPT-4), generating every hour millions of texts, conversations, poems, songs, articles, essays, summaries, reports, recipes, tips, opinions or even code in the desired style and language; consuming, in the process, huge amounts of energy and allowing a massive democratization in the access and personalized generation of information.

For the first time, what we read and especially what we will read is not and will not be the result of the effort of a human, but will be generated by AI systems that have been trained with all the texts ever written. Texts initially created by humans, which will soon coexist with the potentially billions of texts written by Artificial Intelligence systems. Texts, the latter, not necessarily truthful and possibly biased, replicating or even exacerbating patterns of discrimination and prejudice that exist in society.

For the first time, the cold sweat that invades us in the fear of a blank page will be part of the past. We will only have to ask ChatGPT to write us about the subject in question and, in a polite, diligent and instant way –as a good chatbot – will provide us with the desired text, making the dreaded blank page disappear.

We are facing a technological advance that questions the very essence of our humanity. Science tells us that we are lazy by nature: if someone – in this case, an Artificial Intelligence program – can do something for us, we will let them do it; And we also know that, in nature, what is not used, tends to disappear. Imagining a society in which humans stop writing is difficult to conceive, but it is a plausible future. A future without blank pages.

Facing a blank page can be terrifying, but it is also magical and exciting: it represents a world of possibilities, an infinite space in which we can explore and develop creativity, a door to freedom. And although it is undoubtedly difficult to break the perfect white page, once we begin to translate ideas – our ideas – on paper in the form of words, we feel a unique satisfaction to see something that only existed before in our mind take shape. The blank pages are a reminder that our imagination is infinite and that everything is possible if we have the will to take the first step, to write the first word.

Writing is power, said Margaret Atwood, and Umberto Eco reminds us that “to survive, you need to tell stories.” Writing helped Anne Frank “let go of everything, her sorrows disappear and her courage reborn.” We write because we feel (“words are written tears,” Coelho said); we write because we think; we write because we live (“to taste life twice, in the moment and in retrospect,” said Anais Nin); We write because we are. For more than 6,000 years, we have written to understand the world and understand each other, to share, to persuade, inspire or fall in love, to educate, to excite, to empathize, to heal, to remember or to forget…

Therefore, I hope and wish that our future will continue to be full of blank pages that provoke us, intimidate us, inspire us and make us to think. I invite you to join our cause, which is none other than to demand and work to achieve Artificial Intelligence by and for people and the planet. An Artificial Intelligence that really helps and complements us, not that replaces us. An Artificial Intelligence that contributes to progress, a progress, which should undoubtedly include blank pages.