Artificial intelligence (AI) is revolutionizing the relationship between humans and machines with its ability to manage and generate knowledge. It has become a key tool for academic research and there are many Industrial Doctorates projects working in this field. But the generation of artificial knowledge still depends on human beings. Therefore, it is necessary to explain well what is the use that researchers can make of AI to exploit its potential intelligently and ethically.
Artificial intelligence (AI) is on everyone's lips, it seems evident that we are facing an unprecedented technological revolution. The relationship between humans and machines is at a turning point, representing an exponential leap in terms of access and use of knowledge. Elisenda Bou, a leading figure in AI in Spain, who has directed four industrial doctorates, is convinced that one of the most subtle virtues of AI is its ability to humanise machine interfaces: "artificial intelligence makes more people have access to technology"
It should be remembered that, today, artificial intelligence has the capacity to manage human knowledge, but also to generate knowledge. Very common techniques in the field of research such as Machine Learning, Big Data or Deep Learning have created many possibilities for artificial intelligence and computer science in general. Thanks to these and other technologies, AI can manage and process data in massive quantities, which is then also capable of analyzing and extracting very useful information for researchers. Simply put, these combined technologies enable research projects to make discoveries and make decisions based on the knowledge AI provides them.
In fact, AI is helping to make great progress in the creation of human knowledge thanks to machine learning technologies or the creation of predictive models from data. To give an example, one of the fields where the application of AI can generate more benefits in society is the world of health. The analysis of patient data and the creation of predictive models in health are key contributions of the techniques mentioned in the previous paragraph. With these AI techniques, researchers are able to develop predictive models to predict the probability that a patient will develop diseases or disorders, or the reaction they may have to a certain treatment. Without forgetting the ability of techniques such as Big Data to identify patterns and trends that are not obvious to humans, making these decisions with knowledge and being able to provide personalized treatments.
But beyond managing and processing human knowledge, can AI generate "artificial" knowledge? The answer is yes, but with nuances (human, of course). Machine learning is a technique that allows AI to analyze big data and then identify patterns and relationships between these data, which can apparently be hidden from a human being. Discovering these connections between data can allow AI to generate new hypotheses, which leads us to the creation of "new knowledge", and ultimately to scientific discoveries, which in turn can represent innovations for the economic world. We must also consider the numerous ongoing research in the field of creative AI, many of which have already become commercial applications. This specific research seeks the creation of works of art, music and other forms of artistic expression through AI-based technology, obtaining sufficiently significant results that have been compared with works of human creation.
"Beyond managing and processing human knowledge, can AI generate 'artificial' knowledge? The answer is yes, but with nuances."
Creation of artificial knowledge?
But, and now is the time to qualify, the truth is that AI can generate artificial knowledge, but for now it still depends on humans. It is humans who provide the data and, ultimately, who interpret the results. With their capacity for observation, reflection, interpretation and critical capacity, human beings have been generating knowledge for centuries. This human knowledge is very different from the knowledge that a machine can generate. The first is the result of experience, learning and the ability to reason, without forgetting the external factors that influence the whole process (social and cultural context, prejudices, emotions, etc.). On the other hand, the knowledge that a machine can generate is the result of data processed by algorithms and mathematical models. As long as machines continue to depend on humans for data from reality, their ability to generate artificial knowledge will be limited. First basic consideration of the revolution we are experiencing.
In fact, one of the basic functions of research is the generation of knowledge, regardless of whether this knowledge is transferred beyond the source that generates it. It is worth remembering that the Industrial Doctorates Plan is based on this process. The aim of the Industrial Doctorates is to promote the transfer of knowledge generated by research through collaboration between universities and companies. Thanks to this programme, doctoral students apply their knowledge and research results to real problems. A process that contributes positively to the transfer of technology and knowledge to companies and society in general.
As we explained above, the benefits of AI in academic research are diverse and some very relevant. These range from the ability to process large amounts of data in real time to the ability to generate predictive models and hypotheses that can be tested empirically, to the identification of patterns and correlations that can be difficult for a human to identify. There are many Industrial Doctorates projects that base their research on AI and the different techniques it covers. As an example, we can talk about projects on artificial intelligence as a tool to support the diagnosis of injuries or ulcers, projects that apply AI on energy consumption data with the aim of finding consumption patterns and creating models to improve efficiency and reduce energy consumption. Other areas of knowledge with projects that use artificial intelligence as a tool for detecting suicidal behaviour in adolescents, or the application of artificial intelligence to the creative process of advertising agencies, or the application of artificial intelligence for optimal and autonomous control of industrial environments. The possibilities are many, and with the recent emergence of artificial assistant applications of natural language assistants (what we call AI chats) the number of research projects in the field of AI is likely to increase.
Among the hundreds of tools available, ranging from text and image generation to dialogue-focused AI bot applications, we have ChatGPT. He is one of the most famous natural language assistants at this point, not the only one but the most accessible and with a lot of potential. This application is based on language models created by the artificial intelligence research laboratory OpenAI. This American laboratory is made up of a non-profit organization and a commercial subsidiary.
ChatGPT is a natural language model that relies on artificial neural networks and uses "deep learning" techniques to process large amounts of data, learning underlying patterns and trends. The algorithms that make up these models have the ability to generate intelligent answers to user questions, among other tasks related to natural language. The uses of this technology in different languages are many, among which we can highlight machine translation, correction or generation of texts, summarizing or schematizing, among other applications related to natural language processing.
"We must put aside the anguish that the disruptive emergence of this technology is generating in the world of education, higher education and the world of research, to focus on its potential and practical learning"
How can this tool be a critical resource for research?
Definitely making good use of it, exploiting its full potential ethically and responsibly. We must put aside the anguish that the disruptive emergence of this technology is generating in the world of education, higher education and the world of research, to focus on its potential and practical learning. First of all, its most visible uses for research staff revolve around the creation of sketches, or the structuring of texts, improving in any case the learning processes of the research teams. While the entire educational community has reason to be concerned if their students use ChatGPT to copy on their exams or do assignments ( tools exist to detect AI-generated text), what is important is to make the effort to constructively address the use of AI in academia. The same thing happens in the field of research, where doubt is often how this technology will affect the writing skills of doctoral students.
There are many ways in which AI can help academic research. We present below some of those proposed by the ChatGPT application answering the question "How can ChatGPT help academic research?":
- Identification of research topics: ChatGPT can be used to identify emerging research topics and areas of research that need the most attention.
- Generation of research questions: ChatGPT can generate research questions from keywords or phrases related to the research topic. This can be useful for researchers who need ideas for their projects.
- Summary of the literature: ChatGPT can be trained to summarize existing literature on a particular topic. This can help researchers save time on reviewing documents and get a quick overview of what's already been published.
- Data extraction: ChatGPT can also be used to extract data from articles and other research-related sources. This can save researchers time, as they won't have to read all documents manually to get the information they need.
- Data collection and analysis: ChatGPT can help analyze large amounts of data and find patterns and trends. This can be useful for researchers to gain a greater understanding of their datasets and get ideas for future research.
- Generation of hypotheses and ideas: ChatGPT can be used to generate new hypotheses or ideas for research. The model can be trained with data related to a specific topic and then generate suggestions for future research.
- Analysis and revision of texts: ChatGPT can be used to analyze and review scholarly texts, such as articles and dissertations, to find errors and suggest improvements.
In this field it is worth following the contributions that Mushtaq Bilal has made on social networks. This researcher from Syddansk University in Denmark has made a thread on Twitter about how to use ChatGPT intelligently for academic purposes. In summary, the researcher invites the academic community to learn how to combine artificial intelligence with natural intelligence, and to make ethical use of ChatGPT to generate structures that can then be converted into academic texts. In no case to create original content, since applications like ChatGPT use prediction models, so the content is always predictable.
In other words, the responses generated by a chat like this are consistent because AI learns to predict words or phrases in a text or conversation. The text generated by ChatGPT is original content to some extent, as it is based on data it has "learned" while interacting with users. It can make unprogrammed answers, it is true, but AI does not have the creative capacity to generate knowledge from scratch and much less understand the answer, it lacks something basic: awareness.
Although we have presented some of the ways in which research can benefit from the use of AI, these uses must always be qualified in order for ethical research to be carried out. We need to see applications like ChatGPT as a tool and not a substitute for the creative ability to generate knowledge from human beings. There is no doubt that there are many benefits of AI in academic research and its importance for business innovation. But the challenges and implications of its use in academic research, such as the reliability of results, must be addressed.
In conclusion, the academic world and the economic world have a challenge regarding the possibilities of artificial intelligence, promoting its use for more effective, innovative and ethical research.
