AI: Artificial Intelligence

AI Ethics much like AI itself is a relatively new field. The Alan Turning Institute provides a concrete definition:

"AI ethics is a set of values, principles, and techniques that employ widely accepted standards of right and wrong to guide moral conduct in the development and use of AI technologies."

Simply put, AI Ethics is making sure AI is held to the same ethical standards as any other field to ensure the rights and safety of all involved. AI Ethical concerns cover a wide range of issues such as

• AI Biases
• Environmental Impact
• AI Hallucinations
• Ethical Labor
• Privacy concerns

AI tools can generate data that looks highly convincing to the untrained eye. However, It is important to remember that these tool don't show the processes they use to create content or show the sources of their content.  Generative AI tools reflect the biases present in their training data. These biases can originate from various sources including: data inputters, anyone providing data or content (personal bias), the origin of the data (machine bias), and the exclusion of underrepresented or marginalized communities (selection bias). Users can also inadvertently reinforce biases in AI by rephrasing prompts until they receive the answer they most desire (confirmation bias). Generative AI tools amplify and reinforce these biases, and it is crucial to remain critical of outputs/responses.

Otherwise, a growing sense of comfort, acceptance, and trust in what these tools provide can transform into depending on the machine to provide answers, realizations, and/or decisions (automation bias). Refer to academic resources to verify information and analyze the company’s data collection policies or procedures.

Digital Humanities Librarian Mary Ton provides an example of ChatGPT 3.5 bias in her Make my emails sound more masculine experiment conducted on June 1, 2023.

When utilizing AI tools and Assistants consider the following:

• What datasets were used to train the AI tool?
• Are the outputs valid & reliable? 
• Are the results/output reproducible? How would you know?
• Are the sources cited in the generated text real, or are they "hallucinations" created by the tools themselves? 
• How might the outputs be biased, either intentionally or unintentionally?

It's important to consider the impact that AI use has on individuals, society, and the environment. Current AI technology requires an large amount of natural resources to sustain functionality for she sheer amount of data and users it manages. Data centers support the training and maintenance of AI models through facility and IT infrastructure as well as access to powerful energy resources (hydro, nuclear, wind, etc.). These centers also require additional maintenance and assets, such as cooling towers or power grids. Due to how interconnected these environmental resources are to AI model investments, existing concerns over greenhouse gas (GHG) emissions, water usage, and land equity have grown in the greater discussion of this technology’s impact and ethical implications.

Although these issues are not unique to this technology, the speed of deployment and demand partnered with the other ethical concerns have made environmental impact a leading point of discussion.

Large language models, or LLMs, behind Generative AI are trained on massive amounts of data to find patterns. Despite the verbiage used to discuss these tools AI's don't actually think as humans do. They use patterns to predict words images and sounds to then generate new content. While the AI might present this information as though it is factual, it is actually a misrepresentation of a recognized pattern from a vast amount of data. The presentation of these hallucinations can be difficult to identify. For example, a common AI hallucination in higher education happens when users prompt text tools like ChatGPT or Gemini to cite references or peer-reviewed sources. These AIs will often hallucinate content, articles, and even individuals that do not actually exist.

Examples of Hallucinations

• Learn more about hallucinations in Mistrust and Verify on MediaSpace.
• Example from David Smerdon: “A Theory of Economic History” by Douglass North and Robert Thomas
• Here’s why AI Has trouble with hands by Vox on Youtube
• Harvard Business Review article (September 2023), "Eliminating Algorithmic Bias Is Just the Beginning of Equitable AI"
• Humanities & Social Sciences Communications article (September 2023), "Ethics and discrimination in artificial intelligence-enabled recruitment practices"
• Scientific Reviews article (October 2023), "Humans inherit artificial intelligence biases"

If a tool uses web scraping, it is collecting public information from the Internet. This data is gathered from anyone who has posted content online or has had their content published online, often without the creators knowledge or consent. Training AI materials on copyrighted material are protected by fair use but exploit artists and authors while also devaluing their labor.

In addition to these concerns, there are also repercussions to workers within the industry. A content aggregator, or moderator, is an individual or organization that collects data. Content aggregators are often employees who train and improve the tool's algorithms. However, some worker communities have been exploited. These employees, often noted as "invisible workers" or "ghost workers," can range from those who train and annotate or label the data to those who enhance and test the algorithm or the models as well as other tasks. Outsourced and contract data workers are especially susceptible to these conditions.

Many companies have hired, or are hiring, professionals to create original content and provide oversight, although this process is still under development.

Examples of Labor Concerns

• Bloomberg article (February 2024), “AI Is Driving More Layoffs Than Companies Want to Admit”
• CNBC (October 2023), “While ChatGPT stokes fears of mass layoffs, new jobs are being spawned to review AI”
• The Washington Post (August 2023), “Behind the AI boom, an army of overseas workers in ‘digital sweatshops’”
• Fortune magazine’s article (July 2023), “Google’s Bard A.I. chatbot is being trained by contractors who say they’re overworked, underpaid, stressed and scared”
• The Wall Street Journal (July 2023), “Cleaning Up ChatGPT Takes Heavy Toll on Human Workers”
• NBC news article (July 2023), “Actors vs. AI: Strike brings focus to emerging use of advanced tech”

Content and prompts submitted into a generative AI tool are often not private, and can lead to significant breaches of data and security. Data gathered by these tools can be shared with the tool’s training dataset and shared in some manner with other users. Many policies share that data is automatically collected to enhance their training data, algorithm, and products (text, graphics, visual, audio, etc.). As with all terms and conditions, a generative AI tool’s privacy policy can be updated anytime. Check the data privacy policy of the tool regularly for how data is collected, used, and stored. It is best practice not to share sensitive information, personal or confidential, of any kind.

A few questions to consider:

• Is the data privacy policy findable?

• Are you asked for consent to provide data or are you automatically opt-in?

• How long is your data held (data retention)?

• What securities are there for de-identification?

• How are privacy risks addressed?

• Is there a third-party company involved? If so, what access and control does this company have over the data collected?