Visual Literacy

Previously, when people talked about AI tools they meant machine-learning models that could learn and make predictions based on data. Examples include an X-ray showing signs of a tumor or predicting if a borrower is likely to default on a loan. Generative artificial intelligence goes beyond predicting. Generative AI tools analyze vast amounts of data to understand underlying patterns and relationships. They then generate new content that resembles the original data. Their primary output is new data such as text, images, music, and videos.  

Examples of how generative AI can be used:

• Generative design software
• 3D modeling tools
• Visualization applications

According to the National Library of Medicine’s definition of generative artificial intelligence (AI), generative AI is “a system of algorithms or computer processes that can create novel output in text, images or other media based on user prompts. These systems are created by programmers who train them on large sets of data. The AI learns by finding patterns in the data and can then provide novel outputs to users' queries based on its findings.”

Text-based tools

• ChatGPT
• Microsoft Copilot

Image-based tools

• ChatGPT
• Canva
• Craiyon
• DALL·E 
• FLUX.1
• Google Gemini
• Ideogram
• Midjourney
• Reve
• Stable Diffusion

For more tools, see Area 49 Resources.

• Authorship & Intellectual Property:
  • AI-generated content ownership is unclear which leads to concerns & disputes.
• Ethical AI Use in Design & Research:
  • Ethical implications of using & sharing content
  • Impact of AI-assisted design on traditional creative processes
  • Environmental & labor concerns
• Flawed source material/data
  • Limited or biased source material can (1) lead to misinformation and (2) influence design, aesthetics, historical accuracy
  • AI models reflect biases of training data
• Hallucinations
  • AI models may generate information that is untrue, such as inventing citations, legal cases, or other details that do not exist. 
  • While hallucinations are less frequent, they still exist.

Which image is AI generated and which is a photograph?

Answer: 

1. AI generated image created in Canva with the prompt "Create an image of the flower Cosmos bipinnatus ‘Sonata White’ in a British garden"

2. Photograph labeled "Cosmos bipinnatus ‘Sonata White’ (cosmos)" from the Shakespeare Garden at the Brooklyn Botanic Garden. The photographer is Alvina Lai.

3. AI generated image created in Google Gemini with the prompt "Create an image of a cosmos sonata white flower in a garden"

Several new sources pointed to AI mistakes in the wedding photos of Jeff Bezos. Two can be spotted in the photograph below from Hindustan Times. 

The image of the left points to the shortened chair leg, and the image of the right shows warped and missing fingers. 

Tech influencer and founder of Safe Futures Online Allison McSorley also pointed out the AI mistakes using the photos below from her website. 

*Disclaimer: Jeff Bezos is a leading investor in AI data company Toloka

Below is a list of things to look for to determine if an image or video is created by AI. These are some indicators that will help you decide.

1. Study the faces, especially the eyes and facial expressions
2. Pay attention to hands. Do the hands looked warped? Current AI struggles to render the complexity of hand movements. 
3. Look for actions that don't make sense. For example, a cat whose whiskers appear and disappear. 
4. Closely look at the objects in the photo or video. Are two objects trying to occupy the same space? Are the tables missing legs? Is the object someone is holding melting into their hand?

For videos: 

1. Check the clarity of the video. Is the video strangely blurry video even though it says it was taken with higher-quality equipment?
2. Watch the lips as the person is talking. Does the movement of their lips match up with the words and sounds?

See more tips here.