Kapitola 1: Základní AI koncepty

How AI works 

Different types of AI rely on distinct algorithms, each tailored to specific tasks. 

One of the most diffused is Language AI. It generates human-like responses by predicting the most probable next word or sequence of words based on the input context. This process involves sampling from a probability distribution over the model's vocabulary or using techniques like beam search to find the most likely sequence of words.

Language AI models typically use deep neural networks (such as transformers), known for their efficiency in handling sequential data, making them suitable for Natural Language Processing (NLP) tasks.

The model is trained using a process called supervised learning. During training, it learns to predict the subsequent word in a sequence given the preceding words. This iterative process occurs over millions of text samples, adjusting the model's parameters to minimise prediction errors. 

Attention mechanisms enable the model to focus on different parts of the input text, facilitating a better understanding of context and capturing long-range dependencies. 

Additionally, pretrained language models are often fine-tuned on specific tasks or domains to improve their performance in applications like text generation, summarization, or question answering. Transfer learning allows models to use knowledge learned from one task/domain to excel in another with minimal additional training.  

Another example can be represented by generative AI. Generative AI harnesses the power of deep learning and neural networks to learn the underlying patterns and distributions in training data and generate new content that is similar to what humans produce. Through training, optimization, and evaluation, generative models can create realistic and diverse samples across various domains, unlocking a wide range of creative and practical applications.

Working on and with AI in schools is crucial for several reasons. 

Education on AI 

It is necessary to educate people to recognise the benefits of AI, but above all, to use AI consciously.

Working on AI will prepare students for a tech-driven future, equipping them with essential skills for the modern workforce. 

Understanding AI ethics and implications fosters responsible use and critical thinking among students, ensuring they navigate technology ethically and responsibly in their future endeavours. Overall, integrating AI into education fosters innovation, enhances learning outcomes, and empowers students to thrive in the digital age.

Efforts should be made to create a culture on AI, making students (and not only) reflect on some fundamental points:

●        Who is responsible for AI choices?

●        Designing simple AI systems: learning to build and train AI algorithms could be a way to use commercial systems of this type consciously.

●        Understanding AI systems: AI, if used correctly, enhances human intelligence! People need to be educated in the correct use, and appropriate skills need to be developed for a life that is truly improved by AI.

Education with AI 

AI enhances learning experiences through personalised education, adaptive tutoring, and interactive tools, catering to diverse learning styles. Moreover, AI streamlines administrative tasks, freeing up educators' time for focused teaching and mentorship.

By creating appropriate intelligent systems, paths for students with different levels and different learning speeds can be differentiated. We could even suggest different paths based on different learning styles; for example, Papert and Turkle in the article "Epistemological Pluralism"[1] propose two approaches to problem-solving (specifically referring to programming problems): hard and soft. These approaches highlight the importance of catering to different cognitive styles in education.

LLM systems, leveraging their advanced natural language processing capabilities, present a promising avenue as tutors for learning. Their ability to comprehend and generate human-like text enables them to provide personalised and adaptive learning experiences tailored to individual needs. LLMs can offer instant feedback, explanations, and clarification on complex concepts, fostering deeper understanding. Moreover, their vast knowledge base allows them to curate diverse learning materials, from textbooks to scholarly articles, enriching the educational experience. Through interactive dialogues and simulations, LLM tutors can engage learners in immersive and interactive learning activities, enhancing retention and comprehension. As these systems continue to evolve, they hold immense potential in revolutionising education by offering accessible, scalable, and effective learning support.

However, it is crucial to recognise that LLM systems come with risks such as biases and over-reliance. Therefore, students need critical thinking skills and guidance to use them effectively and responsibly.