Generative Artificial Intelligence, what it is and how it works

Generative AI is perhaps the most well-known among all existing types, probably due to the widely renowned OpenAI’s chatbot, ChatGPT. It’s a type of AI capable of creating new content, functioning based on Deep Learning. But what exactly is Generative Artificial Intelligence, and how does it work?

What is Deep Learning exactly? How does it work?

This type of learning has been in use since 2006 and has been striving to “emulate” the natural learning of a brain from its inception. Today, it’s indispensable for the development of personalized AI systems and continues to make significant advancements in each of its application fields. One of these fields (getting back to the main topic) is Generative AI.

From its beginnings as a research project to now being a specialized company in robotic products for rehabilitation, Inrobics has maintained a constant in technology development: the use of Artificial Intelligence (AI). Our developers breathe life into robots through it. But now, we’ve expanded the depth of interaction thanks to Generative Artificial Intelligence (GAI), the protagonist of this entry. More on that later…

If you want to learn more about Generative AI, how it works, and its utilities, keep reading…

What is Generative Artificial Intelligence? How does it work?

The functioning of GAI involves collecting thousands and thousands of data/information points that the system (through algorithms) utilizes to generate ideas, also incorporating all the information users provide to the system. This aids the machine in becoming more precise, thus achieving human-like thinking. Generating ideas as a brain would, hence its current ethical and security implications. It enriches any field (personal or professional) and helps imbue depth and personality into machines like our robots.

Moreover, the AI models corresponding to this category can generate new content because they learn and analyze patterns or the structure of previously input data and create content with a similar structure. Hence, the expressions of ChatGPT or the images created from instructions we provide (these inputs are called prompts) seem so realistic to us.

One of the peculiarities of this type of AI is the level of customization we seek with its use. Developers introduce different layers of specific data for each GAI case. To find an experience tailored to our needs, we simply need to correctly indicate what we want, and GAI programs respond to our requests.

Generative Adversarial Networks (GANs), what are they and why are they related to Artificial Intelligence?

Another peculiarity in the functioning of GAI is the use of algorithms called Generative Adversarial Networks, also known as GANs (sounds technical, but we promise it’s easy to understand). Currently, most of the applications we use in our daily lives belong to generative artificial intelligences of unsupervised learning. GANs are nothing more than two opposing neural networks competing in a zero-sum game.

But what are zero-sum games? The operation of Generative Artificial Intelligence

They’re a category of games in which one player’s gains balance with another player’s losses. That is, the total losses and gains of the game always equal zero, creating a perfect equilibrium.

Anyway, back to GANs, the first network, the generative one, produces the initial data as requested, while the second network, the discriminative one, identifies and analyzes the material produced by the first and decides if it conforms to the specified requirements or if it belongs to the dataset previously introduced. These networks are the key to the customization and adaptation that GAI applications give to the instructions we ask for, whether in text, video, audio, or image format, and they can do so instantly and without any human supervision.

In fact, despite appearing instantaneous to us, this process involves thousands of showdowns between the generative and discriminative networks. One provides material, the other indicates the degree of correctness of this material (with respect to the stored data from previous training), generating a new attempt by the first, and so on repeatedly. Each time, the generative network learns from mistakes, becoming more precise and quicker at guessing.

These are some sectors where the use of Generative AI is most relevant:

The media buzz surrounding Artificial Intelligence in general, and specifically around GAI, is impressive, directly proportional to the increasing use of tools applying this technology in thousands of fields, both personal and professional.

Technology and Software:

GAI is used to generate code, streamline user interface design, and enhance efficiency in developing autonomous algorithms. It’s an excellent resource for speeding up programmers’ creation processes and correcting their work.


This is a sector we’ve extensively discussed on our blog, along with the most common applications of AI and how clinical professionals use it. Some of the standout uses include assisting in medical diagnosis, automatically generating medical reports, aiding in biomedical research by creating predictive models, comparing and contrasting images, personalized medicine (AI analyzes specific patient data, such as genetics, medical history, and lifestyle, providing a new perspective for specialists to interpret at their discretion), virtual health assistants, or augmented reality in surgery, among many others.

Art and Creativity:

In this realm, it’s used for generating music, graphic design, visual art, and creative writing, offering new perspectives and countless artistic possibilities. This field sparks much controversy because, with AI’s increasing precision, it’s nearly impossible to distinguish between works and writings created by human hands and those artificially created. In this regard, professionals’ responsibility to use these applications as work tools to generate ideas and not start from scratch plays a fundamental role. An idea, a text, or a musical piece created by AI can be infinitely improvable or adaptable and can even inspire the creation of a work 100% elaborated, composed, or written by a human author.

Advertising and Marketing:

In this case, GAI is employed to personalize advertising content, analyze market data and predict trends, create more effective ads, and overall streamline the entire creation and analysis process.

Human Resources:

You may have heard this already, and some may think it’s a myth, but GAI is applied in selecting profiles for job vacancies in companies. Again, under professional supervision or for the automated evaluation of candidates, at least initially where a series of filters are applied to discard profiles and reduce the candidates subsequently reviewed by selection teams.

Data Science:

This is where the training data we mentioned earlier comes from. The data provided to generative artificial intelligences to make them more precise. It may sound somewhat convoluted, but often, already developed GAI help generate synthetic data used in machine learning model training, facilitating algorithm development and testing.

Applications of GAI

Among the most common applications of GAI are: writing (in all its variants, styles, and tones), image generation, music and video generation, sound pieces, synthetic audio or voice (as in the case of automatic narrations and readings), simulation and Virtual Reality (VR), data translation and interpretation, among others.

The BIG NEWS, Generative Artificial Intelligence Arrives at Inrobics

Our robots have their own voice

This is when we drop the bomb in this article. The intention of specifying what GAI consists of was not only to inform but also to announce that in recent months, our development team has been working on introducing this technology into our products (our robots!). This implies robots with greater complexity and a more developed personality, if possible.

In other words, with this integration, Inrobics takes a step toward adapting the most coveted technologies. Toward a robotics infinitely more personalized and adapted to human relationships. This evolution allows us to take human-robot interaction to a completely new level. All this, considering that our robots already have an acquired personality and can now interact verbally.

Inrobics’ vision is to create products that impact people’s quality of life, building a more inclusive future. This update brings us closer to that goal.

We invite you to learn more about us. You can get to know our team or get in touch for any comments, interests, or suggestions through our form. Of course, if you want to see our solutions firsthand, you can request a demo – it’s free!

Picture of Ana Albendea

Ana Albendea

Journalist and audiovisual communicator from Rey Juan Carlos University of Madrid, with experience in press and online media (culture and technology). Specialized in Corporate Communication and Advertising through a Master's degree at Complutense University of Madrid. She has 3 years of experience researching and creating content about technology, artificial intelligence (AI), and robotics. She applies her expertise and knowledge in the AI sector to the healthcare industry and its professional audiences. Notable for her commitment and passion for storytelling, her current professional trajectory focuses on showcasing the potential of robotics, specifically social robotics in the healthcare sector. Creativity in driving and communicating the company's mission, which is to improve people’s quality of life and generate a positive impact on society.