As of 2025, this article provides some answers to the following questions: What can AI be used for? How can AI support artistic practice? How can prompts be made more effective?
Artificial intelligence is one of the major forces driving societal change in our time. In particular, large language models (LLMs) based on neural networks, including ChatGPT, have increased awareness of AI’s capabilities as they have become part of everyday life. Applications such as ChatGPT are often referred to as generative AI because they can generate new content – texts, images, videos or audio – based on their training data. This training data consists largely of human-produced content available online. Using it, AI can calculate probable outcomes, such as how a piece of text might continue. These results are constantly evolving through various forms of machine learning. It is important to note that AI has influenced our lives long before the rise of generative AI applications. For example, Spotify’s and Netflix’s recommendation systems, or mobile phones’ virtual assistants like Siri, are based on AI.
In this article, I will outline some general ways in which AI can be used – particularly from the perspective of artistic work. It should be noted that since AI is developing rapidly, the goal is to provide general insight. You can usually find the best solutions for your own needs by exploring the possibilities offered by AI. However, this article aims to highlight key aspects of how AI works and to support AI literacy.
More effective prompts
When working with AI, your primary role is often prompting. Prompts are written input that tells the AI what you want it to generate. Prompts are the main way to influence the outcome produced by AI. While technically skilled users can modify open-source AI systems at a deeper level, for most, prompts are the key to success. Prompts can be optimised when you understand a bit about how AI operates. The following examples focus on text generation, but many of the same principles apply to image video, and audio creation. Consider at least the following aspects when you wish to create more effective prompts:
The importance of the context. AI responses are influenced by the context in which the prompt is given (Rettberg 2023). For example, generative AI often refers to earlier parts of the same conversation. If you want to change the topic, start a new, clean chat window.
Wording and precision. The structure and style of your prompt affect the AI responses (Rettberg 2023). If the result is not what you expected, you can revise the prompt and ask again. The more detailed your prompt is, the closer the responses will be to what you want. For example, you could ask for an academic-style response or a simplified explanation suitable for a secondary school student.
Text continuation. AI models generate text based on previous input (Toivonen 2021). If you have a draft or passage you would like the AI to continue, paste it into the chat window and prompt it to continue the text.
Request multiple options. AI generates responses using probability calculations based on its human-produced training data. This often results in generic “average” answers. You can improve the output by asking for multiple options such as 20 different title ideas.
Common topics or requests. AI output is affected by how common the request in the prompt is (Rettberg 2023). This is because the more training data there is available on the topic, the better the AI responses usually are. In the case of literature, for example, AI tends to perform well with genre-specific writing (Bajohr 2022, 274–275). If the topic is obscure, the AI is more likely to produce errors or misleading content, which is sometimes humorously referred to as hallucinations (see Arkoudas 2023).
Politeness or pressure. AI seems to have learned that polite requests generally yield better outcomes (Green 2023). Being polite – asking nicely and saying thank you – can therefore improve the quality of responses. On the other hand, adding pressure can also enhance results. For example, AI may produce better responses if you state in your prompt that everything depends on this response, and that you are relying on the AI to generate its best effort.
What can AI be used for?
The power of artificial intelligence lies in its ability to enhance, or even automate, a wide range of tasks performed by humans. However, AI is not an all-knowing project manager. It requires a goal set by a human, and its outputs must be supervised by a human as well, as the user is ultimately responsible for disseminating any incorrect or misleading content it may produce. In research and development contexts, today’s AI systems are often referred to as “narrow AI” or “weak AI”. This means that AI can simulate certain human abilities but does not possess true intelligence. Whether AI will ever progress to a so-called artificial general intelligence remains highly debated.
In certain clearly defined tasks, AI can outperform humans (Mahowad et al. 2024, 2). It serves best as a powerful assistant that can, at its best, lead the user’s thinking in new directions. To make creative work smoother, it is important to find the right tools for your needs. Applications can also be combined, using one AI application’s output as a starting point for another. Below are some examples of AI applications available in 2025 that support artistic practice:
Text generation and editing. ChatGPT and other large language models can generate text drafts or stories, or answer questions. Grammarly, for example, assists with proofreading and grammar.
Image and video generation/editing. DALL-E and MidJourney generate virtual images based on textual descriptions given in prompts. Synthesia creates videos. Adobe’s programs enable the processing of images and videos using AI.
Music composition. AIVA, Suno, and Amper Music can compose music in various styles.
The following list describes some of the most common uses of AI applications:
Ideation. You can use AI for ideation and further development of your artistic work. For example, you can prompt AI to examine the choice of narrator in a text or to suggest a dialogue that could be developed around a certain topic. According to a report published by Teosto (2025), AI is used in the field of music to create lyrics, seek inspiration and generate ideas. AI can also offer interesting perspectives on various topics, as it is capable of processing vast amounts of data and may uncover something that challenges conventional thinking.
Drafting grant and other application templates. You can ask AI to generate a generic application template, or you can input more specific information about the application target, making your application more personalised. You can refine the application to better reflect your own voice by feeding it your previous applications. In that case, AI will pick up on your characteristic linguistic features. It is essential to remember not to input anything into AI that you would not want to become public. AI can also assist in project ideation, improving portfolios or articulating your skills in different contexts. You can read more about the grant application process here.
Enhancing communication. You can use AI as an assistant in tasks such as generating content for social media, drafting press releases and refining your target audience. You can find extensive information about communication in the arts sector here.
Creating summaries. AI is an effective summariser if you wish to know whether an article is worth reading, for example. AI can also summarise your own texts, including if you need to condense your expertise into an elevator pitch.
Proofreading and finalising artistic work. You can ask AI to check grammar or the structure of a sentence you are unsure about. However, it is worth noting that AI often produces average responses, which means that unique language may be filtered out even if it is grammatically correct. In the case of music tracks, images or videos, AI can also assist with finishing touches and correcting errors. For example, Paul McCartney and Ringo Starr used AI to help finalise the Beatles’ track “Now and Then” based on a demo recorded in 1978 (2023).
Translating texts. AI is also capable of translating texts from one language to another. However, it may not always grasp the full context of a text or the nuances of terminology, so it is important to explain the background in the prompt. You can also predefine any special terms in the text. AI’s translation capabilities vary depending on the language pair, so the quality of the translation should be verified, particularly in formal contexts.
Supporting artistic expression. AI can help guide creative work in new directions by offering artists a kind of new medium – or at least different tools and starting points. For example, in media and video art, AI is used to generate images or create hybrids that combine different artistic genres. In music, AI can be used to develop melodies, harmonies or entirely new soundscapes (Teosto 2025). AI can also be part of text production. For instance, Sean Michaels used AI in his novel Do You Remember Being Born? A Novel (2023) to generate the texts for fictional conversations with AI in the work. In visual art, Sami Lukkarinen’s exhibition Borderland (2023) included paintings based on AI-generated images.
Creating virtual characters. Among other things, AI enables you to create chatbots with which an audience can interact as though they were conversing with a fictional character you have designed. In literature, readers of Niklas Ahnberg’s youth novel Poika joka paransi haavat (2023) can continue exploring the book’s world through virtual conversation. Using the ChatGPT language model, readers have been given the opportunity to speak with a character who has passed on in the story – something Ahnberg hopes will help young readers process the difficult emotions following the character’s death (Publishing Company Aula & Co 2023). In the music world, it is also possible to create virtual musicians. An example is the virtual artist Tello, developed by musician Stig and producer Matias Melleri, who contributes to songwriting, singing and marketing.
Engaging the audience. AI can be used to create works that respond to the actions of the audience. For example, installations can be programmed to react in different ways to sound or movement in the space, which then shapes the work. Examples of such works include I Magma (2019) by Jenna Sutela and Not The Only One (2018) by Stephanie Dinkins, in which AI interacts with the viewer (see Poutiainen 2024, 75). Literary works, in turn, may evolve based on the choices of the reader.
Something completely different… When using AI, it is worth considering what it is capable of. AI continues to evolve and can sometimes exceed its purely instrumental role by producing unpredictable results. AI can therefore be approached not only as a tool but also as a medium that shapes the artist’s perspective – or even as a partner that participates in the construction of a work (Lundman & Nordström 2023, 658). Perhaps, AI can be used to achieve something aesthetically autonomous, something that does not merely recycle human-created content from a new angle.
Quick summary of how to use AI effectively:
AI is a powerful assistant that can automate parts of your work such as writing application drafts and planning communications. It can also rapidly generate large amounts of text, images, video and audio.
It is important to understand a bit about how AI operates to get the best possible response to your prompt. Prompts are often more effective if they describe the context and define the desired result as specifically as possible. Often politeness – or pressure – can help achieve better results.
AI can support artistic expression in many ways from ideation and finalising artistic work to increasing audience interaction or even exploring unknown directions.
Finally, it must be remembered that AI requires human input, as the user is ultimately responsible for the goals pursued.
Sources:
Arkoudas, Konstantine (2023). “ChatGPT is no stochastic parrot. But it also claims that 1 is greater than 1”. https://medium.com/@konstantine_45825/chatgpt-is-no-stochastic-parrot- but-it-also-claims-that-1-is-greater-than-1-e3cd1fc303e0 (accessed 5 March 2025).
Bajohr, Hannes (2022). “The Paradox of Anthroponormative Restriction: Artistic Artificial Intelligence and Literary Writing”. CounterText 8, no. 1 (2022), 262–282.
Green, Ben (2023). “Why saying please to ChatGPT could give better answers”. https://medium.com/ @ benjclaytongreen/why-saying-please-to-chatgpt-could-give-better- answers-b25ca1d7a2ed (accessed 5 March 2025).
Mahowald, Kyle et al. (2024). “Dissociating Language and Thought in Large Language Models”. Trends in Cognitive Sciences, Volume 28, Issue 6, 517–540.
Nordström, Paulina & Lundman, Riina (2023). “Creative Geographies in the Age of AI: Co-creative Spatiality and the Emerging Techno-material Relations Between Artists and Artificial Intelligence”. Transactions of the Institute of British Geographers.
Poutiainen, Viivi (2024). “Oraakkeli, narri ja peili – Tekoälyn roolit taiteessa”. In Hereillä mutta mihin hintaan: Nykytaiteen millenniaaliantologia (eds.) Roosa Kuosmanen, Sanna Lipponen, Anu Pasanen & Viivi Poutiainen. Khaos Publishing, 72–79.
Rettberg, Scott (2023). “Cyborg Authorship: Writing with AI – Part 1: The Trouble(s) with ChatGPT”, Electronic Book Review. https://doi.org/10.7273/5sy5-rx37.
