Technology is fascinating. Science fiction writer Arthur C. Clarke once wrote, “Any sufficiently advanced technology is indistinguishable from magic.”
This is true, but indistinguishable doesn’t mean technology is magic. It has limitations.
That includes artificial intelligence (AI), and AI-driven technology like ChatGPT, which continues to rise in popularity. OpenAI, the company behind it, is valued at about $29 billion as of 2023.
Despite ChatGPT’s capability of generating coherent, essay-like responses – this can be deceptive. “ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers,” according to information from OpenAI, listing examples of the language model-based chatbot’s limitations.
Thus, it can fabricate content that appears plausible but is entirely fictional. Such is a recent June 2023 case in the United States involving attorneys Steven A. Schwartz and Peter LoDuca blaming ChatGPT for fictitious legal research in a court filing.
The case revolves around a lawsuit against Avianca, a Colombian airline, where the lawyers inadvertently included references to non-existent legal precedents and fictional court cases concocted by ChatGPT.
Their intention to employ AI-powered research to strengthen their client's case backfired when they failed to fact-check the suggestions offered by the AI. The outcome? A bizarre filing that baffled a federal judge, leaving a trail of embarrassment, potential sanctions, and a tarnished professional reputation.
Bizarre as it may be, it helps address a couple of critical points when it comes to AI-powered technologies.
First, AI must be treated with caution and vigilance. This may be an oversimplification, but ChatGPT, like many other AI-powered technology, mimics data fed to it such as sounds, images, words, and even numbers. It does not create, nor does it understand the context behind a query. Behind that computation is mathematics.
Furthermore, the incident illuminates the need for a better understanding of how AI functions. In the case of AI chatbots, it may feel like you’re talking to a real person, but you’re not. Imagine interacting with an extremely advanced calculator, or multiple advanced calculators working together.
That’s how ChatGPT works. It responds to the query provided by a user to generate responses, based on large amounts of data provided by the developers or other users.
Nevertheless, this isn’t to undermine the vast potential or even the present capability of AI. It’s a powerful tool. But it is limited, and AI isn’t magic.
Ultimately, a comprehensive understanding of AI's capabilities and limitations is crucial, enabling users to make informed decisions and mitigate the risks of employing AI irresponsibly, while also benefiting from its use.