To build an Artificial General Intelligence (AGI), technology with an intelligence that is at least compatible with that of humans would be needed, which implies creating a mathematical model that would allow the prediction of behavior. As the human neurocognitive system is complex and therefore impossible to translate into mathematical models, it is not possible to create an emulation of human software. These are the reasons an AGI is impossible, according to Barry Smith, Director of the National Center for Ontological Research (NCOR) and a professor at the University of Buffalo. The author of “Why Machines Will Never Rule the World: Artificial Intelligence without Fear” is in Brazil to participate in the ABES Conference 2022, an event promoted by ABES – Brazilian Association of Software companies that discusses the constant changes in the technological landscape. The event can be seen by ABES channel on Youtube.
Smith explains why it is not possible to emulate human feelings and why AI will continue to be more efficient than humans along certain paths, such as pattern recognition, but will never be able to learn about prediction: “AI only works when there is a certain distribution and in simple systems where the prediction can be made via Newtonian mathematics and logic. We are a complex system, where there are new elements, forces, interactions all the time. Each of our 100 billion neurons is a complex system.”
As an example, Smith talks about online translation systems that don't understand the meaning of words. “If you translate something from English to Portuguese, from Portuguese to German and then to English again, there will be no semantics and the sentences will lose their meaning”, he summarized. According to him, these systems use 213 million parameters that will never be “explainable” and create simplified models of languages. “They do a good job, but they fail in front of a human translator, because there are many aspects of linguistics that are not conscious, they have many parameters and variations. The online translator works with what is static and not with cognitive semantics”, explained the professor during the opening of the ABES Conference 2022.
Following, Marc Etienne Ouimette, Global Lead, AWS AI Policy; Aline Oliveira Pezente, Co-founder and Chief of AI & Product Strategy at Traive; and Caio Guimarães, Partner of the Boston Consulting Group GAMMA (Artificial Intelligence division) talked about how they use active intelligence in decision making, the importance and results in ensuring people's data literacy and how to transform AI, Big Data results and Analytics in Actionable Business Intelligence. According to them, active intelligence is not just for large companies, but it is necessary to change culturally so that decisions are made based on data. And for that, hiring and training talent is essential for active intelligence to be real.
The training, retention and attraction of professionals was also a highlight of the second roundtable, which discussed how to have an abundance of talent in a world without borders. For Eva Pereira – Alianças, Marketing e Legal da WOMCY (Women in Cybersecurity), focusing on diversity, equity and softskills can be the key to the growth of talent in a company. Izabel Branco, Vice-President of Human Resources at Totvs believes that in addition to technical skills and business vision, looking at what each one can deliver is essential: “Careers in Y are very important. It is necessary to know how to recognize a person who is technical and does not want to be a manager and, in addition, to recognize what is important for that role, not necessarily for that career”, he explained. On the other hand, David de Oliveira Lemes, Director of the Faculty of Interdisciplinary Studies at PUC-SP and Guilherme Neves Cavalieri, Academic Director of XP Educação defend that technical training combined with the development of soft skills should be a basic premise for the training of students in the technology.
Follow the event via ABES channel on Youtube.
EN 
PT 
