GONZALO RECIO. Master in Artificial Intelligence UPC and coordinator of research and innovation laboratory of the Barcelona School of Informatics (inLab FIB) at the Universitat Politècnica de Catalunya – BarcelonaTech (UPC).
Artificial intelligence has been shown to be capable of creativity by demonstrating great skills in tasks that involve a certain degree of imagination, such as the creation of texts and art. Creativity has always been considered a mainly human characteristic that could never be reproduced by machines or algorithms, at least not in the near future. Nevertheless, today artificial intelligence (AI) and Deep Learning are achieving incredible milestones in this aspect and the time has come to reconsidered whether we have reached the point at which creativity could be automated by machines.
The answer to this question is not easy, as creativity is a complex human trait that involves many cognitive processes, such as ideation, problem solving and imagination. In addition, creativity is often associated with the capacity to conceive new, original ideas, something that AI is still struggling to achieve, and it is still far from being able to replicate human creativity 100%.
In general, AI is very good at understanding and analysing data, but it is not as good at creative tasks. For example, a computer can easily beat a human in a game of chess, but it would be very difficult for it to come up with a new chess strategy. Even so, there are some tasks involving creativity, such as the creation of art texts or images, in which AI has shown great ability. For example, a painting created by an AI algorithm sold at a Christie’s auction for 432,500 dollars. This figure is way above the estimated price of between 10,000 and 30,000 dollars. In addition, there are many examples of texts created by AI through models, such as GPT-3 of OpenAI, which can generate texts that are indistinguishable from those created by humans.
Some experts believe that we have reached the point at which creativity could be automated by machines. However, others believe that there are still some limitations to what AI can do in terms of creativity. For example, AI is still not capable of creating totally new ideas, but it can help humans to develop them by providing data and analysis.
In general, it is clear that AI is capable of being creative today. Nevertheless, it is important to consider that we are still far from being able to replicate the entire scope of human creativity. This is because creativity is a highly complex trait that involves a lot of cognitive processes that are difficult to replicate. However, AI has demonstrated great potential in tasks that require creative thinking and it is likely that in the future it will be able to create ideas that are even more complex and original.
All the text you have read up to this point has been generated by the AI model GPT-3. As input, the model was just given the first paragraph in English. The rest is all text generated by a neural network. Then, the text was translated with DeepL and Google Translate and subsequently corrected in the same way that a human text would have been. The images in the article were generated by Stable Diffusion, a well-known model of deep learning. Therefore, 95% of this article has been automated with artificial intelligence models. This shows that currently AI is capable of carrying out tasks in which creativity is a key factor. In short, AI is talking about itself and creating a self-portrait. This is quite surprising.
AI models such as these have been used in real projects by inLab FIB, the innovation and research laboratory of the Barcelona School of Informatics. Projects include Roses UPC – Sant Jordi Digital 2022, in which users can create and share personalised digital roses with deep learning models, indicating an artistic style and a short description (https://roses-inlab.fib.upc.edu/). As you can see, artificial intelligence opens up infinite new possibilities and applications, which we thought we would see in the distant future but already form part of our present.
“Creative Artificial Intelligence”
“Creative Artificial Intelligence”
Project “Roses UPC – Sant Jordi Digital 2022” developed by inLab FIB (https://inlab.fib.upc.edu/ca/roses-upc-sant-jordi-digital-2022)