Large language models, such as OpenAI’s ChatGPT, have revolutionized the way artificial intelligence interacts with humans, producing text that often seems indistinguishable from human writing. Despite their impressive capabilities, these models are known for generating persistent inaccuracies, often referred to as “AI hallucinations.” However, in a paper published in Ethics and Information Technology, scholars Michael Townsen Hicks, James Humphries, and Joe Slater from the University of Glasgow argue that these inaccuracies are better understood as “bullshit.”
Large language models (LLMs) are sophisticated computer programs designed to generate human-like text. They achieve this by analyzing vast amounts of written material and using statistical techniques to predict the likelihood of a particular word appearing next in a sequence. This process enables them to produce coherent and contextually appropriate responses to a wide range of prompts.
Unlike human brains, which have a variety of goals and behaviors, LLMs have a singular objective: to generate text that closely resembles human language. This means their primary function is to replicate the patterns and structures of human speech and writing, not to understand or convey factual information.
The term “AI hallucination” is used to describe instances when an LLM like ChatGPT produces inaccurate or entirely fabricated information. This term suggests that the AI is experiencing a perceptual error, akin to a human seeing something that isn’t there. However, this metaphor is misleading, according to Hicks and his colleagues, because it implies that the AI has a perspective or an intent to perceive and convey truth, which it does not.
To better understand why these inaccuracies might be better described as bullshit, it is helpful to look at the concept of bullshit as defined by philosopher Harry Frankfurt. In his seminal work, Frankfurt distinguishes bullshit from lying. A liar, according to Frankfurt, knows the truth but deliberately chooses to say something false. In contrast, a bullshitter is indifferent to the truth. The bullshitter’s primary concern is not whether what they are saying is true or false but whether it serves their purpose, often to impress or persuade.
Frankfurt’s concept highlights that bullshit is characterized by a disregard for the truth. The bullshitter does not care about the accuracy of their statements, only that they appear convincing or fit a particular narrative.
The scholars argue that the output of LLMs like ChatGPT fits Frankfurt’s definition of bullshit better than the concept of hallucination. These models do not have an understanding of truth or falsity; they generate text based on patterns in the data they have been trained on, without any intrinsic concern for accuracy. This makes them akin to bullshitters — they produce statements that can sound plausible without any grounding in factual reality.
The distinction is significant because it influences how we understand and address the inaccuracies produced by these models. If we think of these inaccuracies as hallucinations, we might believe that the AI is trying and failing to convey truthful information.
But AI models like ChatGPT do not have beliefs, intentions, or understanding, Hicks and his colleagues explained. They operate purely on statistical patterns derived from their training data.
When they produce incorrect information, it is not due to a deliberate intent to deceive (as in lying) or a faulty perception (as in hallucinating). Rather, it is because they are designed to create text that looks and sounds right without any intrinsic mechanism for ensuring factual accuracy.
“Investors, policymakers, and members of the general public make decisions on how to treat these machines and how to react to them based not on a deep technical understanding of how they work, but on the often metaphorical way in which their abilities and function are communicated,” Hicks and his colleagues concluded. “Calling their mistakes ‘hallucinations’ isn’t harmless: it lends itself to the confusion that the machines are in some way misperceiving but are nonetheless trying to convey something that they believe or have perceived.”
“This, as we’ve argued, is the wrong metaphor. The machines are not trying to communicate something they believe or perceive. Their inaccuracy is not due to misperception or hallucination. As we have pointed out, they are not trying to convey information at all. They are bullshitting.”
“Calling chatbot inaccuracies ‘hallucinations’ feeds in to overblown hype about their abilities among technology cheerleaders, and could lead to unnecessary consternation among the general public. It also suggests solutions to the inaccuracy problems which might not work, and could lead to misguided efforts at AI alignment amongst specialists,” the scholars wrote.
“It can also lead to the wrong attitude towards the machine when it gets things right: the inaccuracies show that it is bullshitting, even when it’s right. Calling these inaccuracies ‘bullshit’ rather than ‘hallucinations’ isn’t just more accurate (as we’ve argued); it’s good science and technology communication in an area that sorely needs it.”
OpenAI, for its part, has said that improving the factual accuracy of ChatGPT is a key goal.
“Improving factual accuracy is a significant focus for OpenAI and many other AI developers, and we’re making progress,” the company wrote in a 2023 blog post. “By leveraging user feedback on ChatGPT outputs that were flagged as incorrect as a main source of data—we have improved the factual accuracy of GPT-4. GPT-4 is 40% more likely to produce factual content than GPT-3.5.”
“When users sign up to use the tool, we strive to be as transparent as possible that ChatGPT may not always be accurate. However, we recognize that there is much more work to do to further reduce the likelihood of hallucinations and to educate the public on the current limitations of these AI tools.”
The paper, “ChatGPT is bullshit,” was published June 8, 2024.
