Artificial intelligence (AI) has recently experienced exponential growth in many areas of knowledge, with the academic field not setting apart from this trendy tool. Most students use AI to make quick queries through ChatGPT or similar systems to resolve doubts or carry out reports or projects on a particular topic. Obviously, from the lecturer's point of view, the use of AI by students to write assignments and projects raises serious doubts about whether it improves their learning capabilities and provides them with the appropriate knowledge and skills required in each subject. Furthermore, the indiscriminate use of AI has caused serious problems with intellectual property.

On the bright side, there are different applications in parallel with using AI as a quick and easy tool to seek information. More precisely, other types of AI can be used in complex learning environments, such as Higher Education. One of the most in-demand types of AI is neural networks. This type of AI is characterised by using layers of nodes that approximately simulate the behaviour of a natural neural network. There are multiple combinations in the number of layers and nodes in each layer. Each combination will work best for a specific type of problem. The artificial neural network needs data to train, as natural networks do. However, they do such a process differently; while humans learn from birth to old age, artificial networks need training data. In such a process, the larger the data, the better. Once trained, the network can perform tests, verifying that the learning process has been satisfactory and that the chosen neural network is adequate to resolve the specific task for which it is used. Such a problem mainly comprises fitting and interpolating.

Neural networks in Higher Education can be used in two activities: lecturing and research. From a lecturing point of view, it is a new and interesting way to teach students data-fitting methods without resorting to traditional Calculus methods in Mathematics. The use of neural networks is not limited to STEM-type subjects but can be used in any subject where the interpolation of a specific value is required. Any study that contains numerical data can use neural networks, for example, in Educational Sciences or Geography. Even studies and subjects with non-numeric data, such as text data, may be susceptible to dealing with neural networks if the text is codified numerically. AI can also enhance university research in general and data fitting in particular. Artificial networks are promising substitutes for traditional statistical tools, including one-way or two-way ANOVA and design of experiments (DOE). Such artificial networks can determine the most significant parameters in processes for prediction and optimisation.

This emerging and trendy tool is thought to play a vital role in the future, making it necessary to make current students aware of the potential of such a technology. This urgent educational need drives this work to explore data fitting with neural networks covering several Chemical Engineering topics. As a result, a comparison will be established between data obtained with traditional Chemical Engineering approaches and those obtained with neuronal networks. This will allow students to understand and gain new and exciting insights into using this technology in the real world, which will help educate the students of the present for future needs and problems they will tackle during their careers.

Keywords: Neural networks, data fitting, interpolation, neural nodes.