sexta-feira, 14 de julho de 2023

Harvard University's ban on "feeding" ChatGPT and the meaning of China´s stellar Pisa results



In a previous post, link above, a reference was made to the recent case of a major US telecommunications company that imposed a ban on its employees from using ChaGPT. The purpose of this restriction was to safeguard the privacy of customer information, as the company recognized the potential risks associated with the "feed" of generative AI tools. Interestingly, Harvard University recently made public a set of guidelines (https://huit.harvard.edu/news/ai-guidelines) regarding the use of generative AI models. These guidelines explicitly prohibit feeding confidential data or non-public research data into those models.

With regard to ChatGPT, it is worth noting that yesterday, the esteemed journal Nature published a study demonstrating that ChatGPT does not offer significant advantages to individuals with strong writing skills. However, it does play a crucial role in reducing the existing inequality between those with weak writing skills and those with greater proficiency. This contributes not only to the democratization of writing ability but also to its devaluation.  https://www.nature.com/articles/d41586-023-02270-9

Asked which areas generative AI will not replace humans, ChatGPT provided the answer reproduced below in blue color. However, it is important to consider that nanotechnology is already operating at a scale much smaller than the "fine motor skills" mentioned in paragraph 4. For instance, chips with dimensions in the range of a few nanometers, thousands of times smaller than the thickness of a human hair, are already being produced. This indicates that the human advantage in the domain of "fine motor skills" may not be long-lasting and will eventually be surpassed by machines. Still, human creativity, emotions, and ethics will continue to be key differentiators. In light of this, it is worth revisiting the content of a 2019 post on the meaning of world leadership for Chinese students on the PISA test, and where it was written that the future of education will have to train innovative humans, with high ethical and moral standardshttps://pacheco-torgal.blogspot.com/2019/12/pisa-best-students-in-world.html

"Although generative AI has made significant advances and has the ability to perform complex tasks, there are areas where humans still play essential roles and will not be easily replaced. Here are some of these areas:

  1. Creativity and Art: While generative AI can generate creative content such as music, paintings, and texts, it often relies on existing patterns and data. Originality, emotional expression, and the ability to transcend conventions are aspects where humans still have an advantage.

  2. Empathy and Human Interaction: Generative AI can simulate conversations and human interactions to some extent, but genuine empathy and emotional understanding are still unique human skills. Human interactions require a deep understanding of emotions, social context, and nuances that AI is still far from matching.

  3. Ethical Decision Making: Generative AI operates based on algorithms and data provided to it, but it lacks the ability to discern complex ethical values or consider the moral and philosophical aspects involved in important decisions. Ethical issues require human judgment, weighing of values, and contextual considerations.

  4. Motor Skills and Physical Manipulation: While generative AI can generate images, videos, and motion simulations, it does not have the ability to manipulate physical objects in the real world with the same dexterity and precision as humans. Tasks that require fine motor skills, such as surgery or assembling complex components, are still dominated by humans.

  5. Creating Deep Emotional Connections: Generative AI can offer assistance and interaction, but it cannot yet replicate the deep emotional connections that humans develop with each other. Personal relationships, genuine empathy, and emotional support are aspects of human interaction that cannot be completely replaced by AI.