Artificial Intelligence (AI) has been advancing rapidly in recent years, and Generative AI is one of the most exciting and quickly evolving fields. Content creators can use this technology to their advantage while minimizing potential drawbacks.
Generative AI is a powerful and rapidly developing field of Artificial Intelligence that has the potential to revolutionize how content is created.
This course will provide a comprehensive overview of current generative AI technologies, with a focus on text and image generation, video creation, and 3D modeling, among other possibilities.
Part I. Introduction.
This course will start with an introduction to the basic concepts of Generative AI, including its application fields, relevant models and architectures, and the tools available for implementation. We will then discuss the many potential business applications for Generative AI, such as automated content generation, translations, sentiment analysis, and summaries. We will also discuss how Generative AI can be used to create images, program code, poetry, and artwork.
Part II. Generative AI models and platforms.
The second part of this course will focus on the current Generative AI models and platforms, from the basics to specific techniques for different types of content. We will also discuss several useful tools, such as Stable Diffusion, DALLE-2, Midjourney, GPT, chatGPT, and a new collection of possibilities. We will explore their main features and possibilities, and how to boost and improve the creative process of artists, writers, and content creators.
Part III. Implications of Generative AI.
Finally, we will look at the implications of Generative AI, both in terms of opportunities, such as the potential of “prompt engineers”, and in terms of potential risks, such as the challenge of ensuring intellectual property rights.
This course is especially aimed at those with an interest in:
- Know the fundamentals that underlie current generative models. Without technical or mathematical depth, but with enough rigor to know its mechanisms.
- Know the different current possibilities in generative models capable of creating high-quality content in different fields.
- A broad vision of the current spectrum of available solutions and being able to take advantage of their advantages and disadvantages.
The course does not require prior technical foundations and is open to:
- Content creators, in the widest possible spectrum.
- Professionals in marketing, communications, art, writing, etc.
- Anyone who wants to streamline and enhance their work, daily activities, or hobbies with the most advanced AI-based generative assistants and techniques of the moment.
Introduction to basic fundamental elements, skipping mathematical background, and any other prerequisite.
Online – Blended learning
- Teaching materials: traditional content + videos
- Asynchronous videos focusing on main concepts, skipping much more dynamic details, following standard recommendations: 7 min. max, main concepts, real teacher on screen, free use of any tool (standard template for slides), screen casts, other possibilities when required.
- Text: Reading material with full explanations, references, links to additional material, and video slides.
- Weekly synchronous (also recorded) sessions of the scheduled topics of the week: introduction to the topic, summarizing and a better way to approach to the work of the week.
- Tutorial support, with online meeting scheduling if required.
- Assessment: Project-based assessment with mandatory online defense (synchronous/asynchronous based on number of students).
Module 1. Introduction to AI and computational creativity
- Definition of Intelligence and Artificial Intelligence
- Introduction to deep learning and neural networks
- From discriminative to generative models
- Is AI able to be creative?
- Traditional computational creativity tools
- Introduction to the new AI Generative principles
- A first approach to AI Generative areas and possibilities
Module 2. AI Generative in Natural Language Processing (NLP) and text-related activities
- Introduction to Natural Language Processing.
- Large Language Models (LLM) (high level introduction to transformers idea).
- Basic training process in LLMs – pros and cons.
- Main current architectures: GPT, BLOOM, LLaMA, OPT etc. and how fine tuning can provide solutions to specific domains and requirements.
- Reinforcement Learning with Human Feedback. Human-in-the-loop and how difficult it is to avoid misinformation etc.
- ChatGPT, Bing Chat: conversational solutions, pros and cons.
- Basic possibilities in NLP using AI Generative models (summarizing, text creation, sentiment analysis, instruct-based activities etc.)
- Prompting engineering: making the most of LLMs.
- LLMs in education: assisting teachers and students. Pros and cons.
- Emergence abilities of LLMs and well-known limitations (hallucination, misinformation etc.)
- Restrictions in reasoning and planning of LLMs.
Module 3. Generative AI and multimedia
- A brief introduction to AutoEncoders, Generative Adversarial Network (GAN), Stable diffusion.
- The training process of huge models – pros and cons.
- Other architectures: CLIPS, BLIP2, Flamingo, etc.
- Not just images: video, animation, music, sound, and multimedia integration.
Module 4. Current AI Generative tools
- Image generation tools: Midjourney, DALL-E, Stable diffusion, Leonardo.ai, different GANs solutions etc.
- Prompting engineering.
- How to work with the latest proposals (Github, Colab, HuggingFace).
- Video, sound and music solutions.
- Multimedia tools: slides, comics, illustrated novels, etc.
Module 5. Ethics, restrictions, plagiarism, creativity limits, regulation, and future of AI Generative tools
- Current state of open litigious.
- Ethical problems and copyright in AI Generative tools.
- New role of artists and content creators.
- Consequences on education.
- Plagiarism: detection and limitations.
- Consequences of AI Generative tools.
- Future trends on AI Generative tools.
To ensure the follow-up of the course, participation in the forum will be required, through questions, giving answers to other classmates or adding information of common interest (such as links to related topics or news etc.). Minimum of 1 monthly contribution.
The final mark will depend entirely on the evaluation of the final project.
Attendance at synchronous classes will not be mandatory. These classes will be recorded and published, so that any student can access the content whenever they want, with the purpose of adapting to the schedule that suits them best.