Why would artificial intelligence create fake sources?
That’s a question I posed to two free programs, ChatGPT and Deep AI. The answers saddened but didn’t surprise me.
First, an explanation of why. I knew about a New York lawyer last year foolishly cited cases he found via ChatGPT and didn’t verify them. They were fake, and the lawyer was sanctioned.
Then I found a letter from a senior lecturer at Clemson University’s English department that one of her students submitted an annotated bibliography that included six journal articles. None of them had URLs. A little research revealed none of them were real, although the names of the journals were.
So, the lesson here is that we as human ghostwriters can use artificial intelligence for such parts of our processes as brainstorming ideas and helping create an outline, but we’d better verify everything artificial intelligence tells us to make sure we’re not lying.
But I wanted to go a step further, so I asked artificial intelligence.
I first queried ChatGPT, and its response began this way: “AI like ChatGPT wouldn’t deliberately create entirely fake sources on its own volition.”
Right. Sure. Whatever you say.
Then it listed four ways it could happen:
Misinterpretation of Instructions or Data: If ChatGPT is asked to generate sources or references, it might misunderstand the request and generate fictitious ones if it doesn’t have the capability to access real databases or sources.
Flawed Training Data: If during its training, ChatGPT was exposed to misinformation or unreliable sources, it might inadvertently generate content based on these flawed inputs.
Malicious Inputs or Manipulation: In some cases, malicious actors might deliberately feed misinformation to an AI like ChatGPT, hoping that it will generate and propagate fake sources or information.
Technical Errors: Sometimes technical errors or bugs in the AI’s programming could lead to the generation of nonsensical or fake sources.
What do all of these have in common? They come down to humans. Once again, it’s up to humans to make sure artificial intelligence is serving our interests. We can’t be passive.
When I asked the same question to Deep AI, it listed five reasons without denying:
Content creation for marketing and advertising: Companies may use AI-generated content to create fake sources to promote their products or services. For example, a company might use AI-generated news articles to create the illusion of popularity or credibility for their brand.
Data augmentation: AI models can generate fake data to augment or enrich existing datasets. This can be useful for training and testing AI models, but it can also lead to the creation of fake sources.
Research and academic purposes: Researchers may use AI-generated content to simulate real-world scenarios, test hypotheses, or explore the boundaries of language generation.
Mischief and entertainment: Some individuals may use AI-generated content for entertainment purposes, such as creating fake news articles or social media posts for fun.
Disinformation and propaganda: Unfortunately, AI-generated fake sources can also be used to spread disinformation, propaganda, or misinformation.
Then it went one step further and added three challenges that arise from artificial-intelligence-created fake sources:
Difficulty in fact-checking: It’s often challenging to distinguish between AI-generated content and genuine sources.
Undermining trust in media: The proliferation of fake sources can erode trust in traditional media and institutions.
Spreading misinformation: AI-generated fake sources can spread false information quickly and widely, potentially causing harm or confusion.
These, of course, are already happening. So once again, it is up to humans to use critical thinking skills—that sadly seem to be withering away in society today—and parse the truth from the fiction.
The choice is ours.
If you would like to comment on this or any other post, feel free to leave a message. Simply click on the Contact tab and use the provided form.
Let's Start A New Project Together
Contact me and we can explore how a ghostwriter or editor can benefit you.
