Skip to Main Content

Generative AI: AI-generated images

AI-generated images

Always check your subject outline and confirm with your lecturer if and how generative AI can be used in your assessment. Using these tools without permission may be considered a breach of Academic Integrity.

Images generated by artificial intelligence pose unique challenges. The ease of image creation through text-to-image software such as Dall-E, Firefly, Midjourney, and Stable Diffusion, make it easier than ever to find and create images but much more difficult to evaluate. This page provides some thoughts on the challenges in this fast changing area of visual literacy.

Refer to ERNI Easy Referencing for guidance on appropriate citation for AI generated images.

AI image detection

New generative AI tools and generative AI detection tools are constantly emerging. These are just a couple that are available as of December 2023:

AI or Not: https://www.aiornot.com/ 
Illuminarty: illuminarty.ai/en/
Was is AI: https://wasitai.com/
 

Ethical considerations

Copyright/Intellectual Property

Copyright and intellectual property laws are constantly having to adapt to new technologies and modes of production and transmission of works. It's difficult for the law to catch up to the rapidly changing information landscape, including information generated by artificial intelligence, and therefore may not cover these issues in detail.

Is AI-generated material covered by copyright?
In Australia copyright is automatically applied for works with a human author. It is possible that AI-generated materials are not protected by copyright, due to the uncertain level of input. 

While AI-generated content may be exempt from copyright, it is still good practice to acknowledge where you retrieved your information and to add the appropriate attribution or citation.

Does copyright cover works from being used to train AI?
Large amounts of information is required to train AI software, usually trawled from the internet and will usually include copyrighted works. What and how this information is gathered is not usually made apparent to users of AI software. 

According to Arts Law:
Under Australian law, it is not always clear whether you need permission from copyright owners to use their work in this way.
Generally, it is an infringement of copyright to digitally reproduce works without permission. Therefore, if you make a digital copy of work (including works that had been available online) to train an AI, this use will require permission from the copyright holder. For example, if you compile your own training data set by scraping images from publicly available websites, you will need permission from rights holder/s.

 

When using AI generating software, you are also contributing to its on-going training. Be sure not breach copyright by uploading materials without permission.

Further info:

Arts Law Centre of Australia 2022, Artificial intelligence (AI) and copyright: Information sheet, viewed 13 December 2023, https://www.artslaw.com.au/information-sheet/artificial-intelligence-ai-and-copyright/ 

Copyright Agency 2023, Copyright Agency position on artificial intelligence (AI), viewed 13 December 2023. www.copyright.com.au/2023/11/copyright-agency-position-on-artificial-intelligence-ai/

Kelly, C 2022, 'Australian artists accuse popular AI imaging app of stealing content, call for stricter copyright laws', The Guardian, 12 December, viewed 13 December 2023, www.theguardian.com/australia-news/2022/dec/12/australian-artists-accuse-popular-ai-imaging-app-of-stealing-content-call-for-stricter-copyright-laws#:~:text=Australian%20artists%20say%20Lensa%2C%20the,up%20with%20AI%2Dgenerated%20art.

The data used to train AI tools is not carefully selected to ensure accuracy; it includes opinion, biases, stereotypes, and discriminatory views. Furthermore, it uses and creates materials that does not capture an entire population, that is, it is not representative.

It is difficult to evaluate a piece of information when you cannot:

  • ascertain the author and their expertise,
  • follow up on where they sourced their information; or
  • perform basic factchecking.

This may change. But for now ask yourself:

  • does this represent a diverse and inclusive worldview?
  • does this perpetuate racist, sexist, or other discriminatory views?

 

To learn more, explore the two visual articles from Bloomberg and Washington Post:

Nicoletti, L & Bass, D n.d., 'Humans are biased. Generative AI is even worse', Bloomberg, viewed 13 December 2023, https://www.bloomberg.com/graphics/2023-generative-ai-bias/

Tiku, N, Schaul, K & Chen, S Y 2023, 'This is how AI image generators see the world', Washington Post, 1 November, viewed 13 December 2023, https://www.washingtonpost.com/technology/interactive/2023/ai-generated-images-bias-racism-sexism-stereotypes/

The positive and negative impact of generative AI is still becoming apparent. There are concerns that generative AI could lead to an increase in:

  • mis- and disinformation, 
  • deepfakes - realistic yet fake audio and video content 
  • cybersecurity threats

 

It's advisable to not provide personal or sensitive information when using AI tools. Like most online tools, they can collect, store, and potentially share your data. The full ramifications of this in the AI world is not fully known.

Generative AI tools also have the potential to collect and analyse large amounts of personal information, making it easier for misuse to occur. 

See the Office of the Victorian's Information Commissioner's article on Artificial Intelligence and Privacy - Issues and Challenges for more.

Art & AI

There are many other ethical and moral concerns around AI generated images. It raises questions such as:

  • Can AI generated content be considered art? Does it need to be human to be art?
  • How does generative AI affect the livelihood of artists?
  • Could an overreliance on AI decrease our ability to be creative, intuitive, and engage in critical thinking? 

There are no clear answers to these broad questions, but it's important to be aware of the growing discussion around these concerns.
Further info:

Gural, N 2023, 'Is AI art real art? Harvard Arts professors ponder the potential of artificial reality', Forbes, 16 August, viewed 13 December 2023, https://www.forbes.com/sites/natashagural/2023/08/16/is-ai-art-real-art-harvard-arts-professors-ponder-the-potential-of-artificial-reality/?sh=322d48112da3

Mineo, L 2023, 'If it wasn’t created by a human artist, is it still art?', The Harvard Gazette, 15 August, viewed 13 December 2023, https://news.harvard.edu/gazette/story/2023/08/is-art-generated-by-artificial-intelligence-real-art/

Shaffi, S 2023, '‘It’s the opposite of art’: why illustrators are furious about AI', The Guardian, 23 January, viewed 13 December 2023, https://www.theguardian.com/artanddesign/2023/jan/23/its-the-opposite-of-art-why-illustrators-are-furious-about-ai