I was hosting some training earlier today on how to use generative AI chatbots (in this case, Microsoft’s Bing Copilot), and a great question from a participant came up:  Can we really trust the citations these chatbots give us in their output? Do these chatbots have mechanisms in place to assure their sources are legitimate? Or can chatbots get confused about what is a genuinely reliable source of information versus, well, internet garbage (for lack of a better term)? We have all seen generative AI, like ChatGPT and DALL-E, take the world by storm, creating everything from quirky poems to stunning artwork. Many of us have used various chatbots to help us out with school and work projects. But with great power comes great controversy, especially when it comes to the legitimacy of its citations. So, let’s dive into this hot topic.

The Rise of Generative AI

Picture this: a world where machines can whip up a novel, compose a symphony, or even generate a meme that goes viral. That’s the magic of generative AI. These models are trained on massive datasets, learning patterns and relationships to create new content that can be eerily human-like. But as with any good sci-fi plot, there’s a twist.

The Citation Controversy

Now, onto the issue in question—citations. In the academic and professional world, citations are like the secret sauce that gives credit where it’s due. But generative AI sometimes messes up this recipe. Here’s why:

• Data Source Ambiguity: Generative AI models are like magpies, collecting shiny bits of data from all over the internet. This makes it tricky to pinpoint the original sources, leading to a murky mess of information.
• Fabricated Citations: Imagine reading a fascinating article only to find out the references are as real as unicorns. Yep, AI can sometimes conjure up fake citations, making the content look credible when it’s not. Talk about a plot twist!
• Intellectual Property Concerns: Using copyrighted material without proper attribution is a big no-no. It’s like borrowing your neighbor’s lawnmower and forgetting to return it—awkward and potentially lawsuit-worthy.

Addressing the Issues

So, how do we untangle this web of citation chaos? Here are some ideas:

• Transparency in Training Data: AI developers should be as transparent as a glass house about their data sources. This helps in verifying the legitimacy of the generated citations and keeps things above board.
• Improved Algorithms: Think of this as giving AI a citation GPS. Better algorithms can help AI navigate the vast sea of information and generate accurate references, reducing the chances of it getting lost and making stuff up.
• Ethical Guidelines: Establishing clear ethical guidelines is like setting the rules for a game. It ensures everyone plays fair and respects intellectual property, making the AI playground a safer place.

And the Saga Continues…

The controversy over generative AI’s citations is a thrilling tale of technology, ethics, and a bit of chaos. As AI continues to evolve, it’s crucial to address these issues to ensure that the benefits of generative AI are realized without compromising the integrity of the content it produces. So, buckle up and stay tuned for the next chapter in the AI adventure! But for now, the answer to the question posed by the trainee this morning as to whether or not to trust the citations the chatbots are giving us?  Well, for now we really can’t. So be sure to check out those citations for yourself and make the human decision as to whether you feel confident using any particular piece of information.