AI output depends entirely on its input, in the form of the prompt it is fed, the dataset used for training, and the engineers who create and develop it. This can result in explicit and implicit bias, both unintentional and intentional.
To train the system, generative AI ingests enormous amounts of training data from across the internet. Using internet content as training data means generative AI can replicate the biases, stereotypes, and hate speech found on the web. Additionally, historic and ongoing gaps in representation (what languages are found on the internet; demographics of people who work in AI) can introduce bias into AI systems and the algorithms that power them.
Learn more:
There are ongoing privacy concerns and uncertainties about how AI systems harvest personal data from users. Some of this personal information, like phone numbers, is voluntarily given by the user. However, additional data, like IP addresses and user activity, are often being collected behind the scenes as well. This data may then be hacked, shared, or sold to third parties without the knowledge or informed consent of the user.
UT's license for Microsoft CoPilot addresses some privacy concerns. User data is neither stored nor used to train the model.
Learn more:
AI is typically associated with virtuality and the cloud, yet these technologies rely on vast physical infrastructures that span the globe and require tremendous amounts of natural resources, including energy, water, and rare earth minerals.
Learn more:
AI still needs human intervention to function properly, but this necessary labor is often hidden. For example, ChatGPT uses prompts entered by users to train its models. Since these prompts are also used to train its subscription model, many consider this unpaid labor.
Academic publishers have struck deals with AI companies to provide access to books and scholarly journals, without necessarily giving notice to authors. Scholars have raised concerns about how their work may be misrepresented or how citation numbers could be impacted. Artists and writers have objected to having their intellectual property scraped from websites to train AI models, without consent or compensation.
Other forms of AI labor, like content moderation, have been outsourced to developing countries where workers are underpaid and subjected to disturbing, violent, and psychologically scarring content.
Learn more:
Deepfakes are videos, images or audio that appear very realistic but are fake. Using AI tools, people can create deep fakes that make it seem like someone has done or said something they have not, with a potentially damaging impact.
Learn more:
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 Generic License.
