In the early 2000s, when AI – Artificial Intelligence was just the name of a Steven Spielberg film, a friendly little Brazilian robot called Ed I was already using technology, then futuristic, to talk virtually with people and clarify doubts on various topics.
Created in 2004 by Conpet, a Petrobras program aimed at the rational use of fuels and energy conservation, the chatbot was developed with technology and support from the São Paulo company InBot.
Using rudiments of AI techniques, such as neural language processing (NLP) and machine learning, Ed, which was an acronym for “Energy and Development”, is cited in the academic world as one of the first Brazilian chatbots to use AInotably in a corporate context of immense visibility.
“When we put it on air, there were more or less two thousand types of responses to different topics. We saw which topics were most talked about by users and, from there, we created answers to questions they didn’t know how to answer,” said InBot’s founder and head of technology, Rodrigo Siqueira, in an interview with Folha de S.Paulo newspaper.
After 12 uninterrupted years of activity, when he answered up to a million questions in a single dayEd was retired in 2016. An “employee” of Petrobrás, the robot worked with a team of experts in programming, computer graphics, linguistics, energy and psychology.
Akinator, the internet genius
![](https://www.cnnbrasil.com.br/wp-content/uploads/sites/12/2024/08/wp10265163-akinator-wallpapers.jpg)
Another icon of the “firewood AI” era, Akinator was developed in 2007 by three French programmers (Jeff Deleau, Arnaud and Olivi), from the company Elokence. Unlike our fellow countryman Ed, the “internet genius” is a game originally released for PC, designed to be played online.
Those interested in challenging Akinator just went to his website, where they could interact freely with the “genius”. The diviner used his own method of questioning to identify the character the user was thinking about.
In Brazil, Akinator became a “national fever” after being shown by the popular presenter MariMoon on the ScrapMTV program in 2008. It was such a success that the company created a Portuguese version of the game. The genius went viral in 2011, when it was launched as a mobile app.
Although you “guess” through questions with basic answers, such as “yes”, “no”, “maybe”, or “I don’t know”, the game used AI algorithms, in addition to a vast database fed with answers from users themselves to be able to “read” their minds, only limiting the possibilities.
Today, Akinator continues to be available on its website and in app stores, and its “magic” still works very well. After all, the only algorithms that can deliver AI to do the work of genius are neural networks, which were not widely available in 2007.
The evolution of artificial intelligence
The enchantment with intelligent robots and telepathic geniuses is not about the AI itself, but rather about the perplexity of being faced with a system that seems to know more than we do. This is no different than when we first used GPT Chat. All those lines with answers pouring out can seem like magic.
Although these pioneering AI programs used extensive databases and decision trees to interact with their users, it was a kind of symbolic AI, following explicit rules programmed directly by humans and restricted in terms of learning and adaptation.
With current technological advances, the focus of AI has shifted to neural networks and deep learning, which means absurd volumes of data, increased computational capacity and advances in learning algorithms. As a result, AI went beyond explicit rules and began to identify, on its own, complex patterns in data.
In this sense, the launch of GPT (Generative Pre-trained Transformer) in 2018 was an inflection point in the history of AI. Based on deep neural networks, these models are trained on vast bodies of text and able to generate human language coherently.
With GPT-3 and GPT-4 versions, AI not only responds, but understands and generates complex texts, creates content and performs complex tasks.
Other new features, such as Gemini (formerly Bard), from Google, and other similar services, combine their language models with other forms of AI, capable of processing different data formats simultaneously, such as text, image and audio.
What is common about Robot Ed, Akinator, ChatGPT and Gemini is that, regardless of the level of sophistication of their algorithms, they deliver “human” responses to their users.
Share: