vineri, 17 martie 2023

ChatGPT reopens the battle for information supremacy

For more than 20 years I have been following very carefully the adaptation of the European press to contemporary technology. In addition to freedom of expression, there is a topic that has been growing in intensity lately. We are talking more and more about the monetization of the media business in the context of the diversification of income sources, most of them adapting to the innovation that has appeared in recent years.


ChatGPT already has a huge societal impact, just as we were impressed by Wikipedia or Google searches back in the day. We talk a lot about Artificial Intelligence, although few realize that we have already been using certain AI technologies for more than 10 years. To give just one notorious example, Google Translate which uses AI algorithms for instant for the translation of texts.


Content production is a fascinating subject, and it is interesting to approach it from a communication angle, so not just from a technical one. The entire dimension of Artificial Intelligence is based on a few clear building blocks. We are primarily talking about databases and their indexing in various matrices, to give ˝signals˝ to adapted receivers. Then we have, and this is practically the technological core, the algorithm, i.e. more understandable to everyone, a program that instructs how to navigate in that database in order to extract clear information. The last part is related to the interface, that is, it is the form of presentation, be it text, graphic, voice, image or video. Even linguistic variants are forms of presentation, given that English is the pivot in such innovations.


At the first presentation of the product, somewhere around the end of 2022, ChatGPT mentioned that it uses certain databases, that is, documents published before the year 2019. Normally, the size of updated databases is a clear direction of action. A company that provides services in this sector scans and indexes an enormous amount of documents.


An exciting part, however, is the focus of a research on a certain predefined database. I understand that such projects exist at the level of the Belgian institutions. Specifically, in addition to identifying a certain number of legal documents from a certain period, on a certain subject, there is now the possibility of having a quick summary of what was written in the thousands of selected pages. The classic procedures of extracting content or creating new paragraphs are combined by an algorithm to give a final form, which is easily accessible to citizens.


On the Internet, we talk a lot about reference sources, that is, a certain content that makes a difference. We now have institutional information, academic research, articles published by prestigious media, and the list goes on. They are basically, you will say, the recommended links on the first page of an online search. Imagine the future though! The information presented according to the formula of a chatbot will come from certain databases, with the indexing planned by certain people, according to certain working hypotheses. A certain algorithm may suggest the text based on certain codes, for example to use only institutional sources or only academic research. In case of clear instructions, some documents, although in the public space, can be ignored or even classified as fake news.


The battle is just beginning. The battle to own the digital BENCHMARK of information on any subject is on. It's basically a battle to create the most credible and visible online space to inform citizens in chatbot format. Everything up to date, in real time, in all the languages of the Earth, free, and easy to access...


Dan LUCA / Brussels

Niciun comentariu: