ChatGPT Reality Check: How ChatGPT Gets It Wrong (And How You can Choose the Right AI)

Everyone knows about ChatGPT at this point, especially students. You might have tried it out for yourself, too. And for good reasons— it can generate text, solve problems, and communicate ideas through responses that sound human-like.

There’s no denying that ChatGPT is capable of many things and is easy to use, but it does come with limitations and ethical implications. For students like me, it’s the potential risk of misinformation and the challenge for academic integrity. Even though ChatGPT operates on billions of data from books, articles, and websites, it’s not what most people would think as an accurate source of information. So does it stand as a good AI tool for studying?

Let’s get into the latest conversation about ChatGPT, and think to yourself— whether it’s time to consider using more reliable alternatives.

ChatGPT is bullshit

In a recently published open-access article, ChatGPT has been claimed as bullshit (yes, you read that right). This claim was coming from the kind of terminology often used to describe ChatGPT's false statements: hallucinations, confabulations, or lies. These labels tend to take away the serious harm and accountability needed for such outputs. That’s why they might be more precisely understood as outright bullshit. I’m surprised that not only is ChatGPT considered bullshit, but the very fact that ‘bullshit’ is indeed a theoretical concept.

That's not the end of it. According to the article, there are two types of bullshit: "soft bullshit", which refers to producing text without any concern for truth, and "hard bullshit", which involves an attempt to deceive the audience about its intention to generate text without regard for truth. What’s clear is that ChatGPT falls into the category of bullshit— at the very least, it is a soft bullshitter. If we look at ChatGPT as possessing agency, it could arguably be labeled as a hard bullshiter too, since it does give the impression of caring about truth.

One example of ChatGPT’s tendency for bullshit is the creation of fabricated reference lists. This can mislead students into thinking they are citing legitimate sources when, in fact, they are referencing non-existent studies or articles. Take this as an example:

Give me a source that talks about Singaporean English.

Goh, C. C. M. (2017). The English language in Singapore: Its features and functions. In The Routledge Handbook of World Englishes (pp. 207-220). Routledge. https://doi.org/10.4324/9781315696760-17

One quick search of the author credentials, title, and even the DOI number would lead you nowhere, even when they look authentic.

Another example is when ChatGPT sometimes delivers answers that are either vague or unverified. This can negatively impact students’ understanding of key concepts for their research projects, assignments, or tests.

Intellecs.ai as a reliable alternative

In an era where various AI tools like ChatGPT become increasingly used for school and beyond, the need for precise and factual information has never been more crucial. As a student in the humanities, truth and honor are values that I have to uphold every time I study, write, and research. This is where Intellecs.ai can come in to play.

Intellecs.ai is a platform made for students that helps them write, research, and take notes powered by AI. Much like ChatGPT, it also features an AI chatbox for quick answers to your questions. But what makes it a reliable alternative to ChatGPT is that it makes sure to integrate legitimate sources into your work, so you’re not putting yourself in danger of plagiarism.

There are two ways Intellecs.ai got your back: the AI Search feature and PDF access component.

With AI Search, you get results from credible sources available on the web. It also includes the clickable links where the information is based, so you can double-check the references and make sure the retrieved content is up to your standards. This is pretty useful when I start with a general idea about something and want to do a deep dive into the topic.

Intellecs.ai also has the option of PDF access, where you can upload your readings, notes, and articles. It reads the entire PDF file for you, allowing you to simply request for key points or a summary based on its content. In this way, you can study and work on something from your own gathered sources. And the best part? You can access this feature for free. No more tedium of going back and forth between a 40-paged PDF and notes, thank me later.

Bottom line: You have the final say in both the choice and outcomes of using AI.

At the end of the day, both ChatGPT and Intellecs.ai serve as tools for our work and learning. How we gather, assesss, and use AI-generated content is up to us. This all starts and depends on an informed perspective on things.

If you prioritize and believe in producing quality work (free from bullshit) always check the facts. Make yourself familiar with proper referencing guidelines, like APA, MLA, and Chicago Style, to establish credibility in writing. Do explore alternatives, such as Intellecs.ai, which can provide not only advanced features but also a greater sense of security in your research and writing.

To help you select an ideal AI tool for your learning, ask yourself these questions:

• Does the AI tool generate answers that are consistent with existing knowledge in the subject?
• Does the AI tool allow me cross-check its responses with reliable sources?
• Does the AI tool encourage original thought and critical thinking?

By the end of this blog, you should think twice before using ChatGPT. Build trust starting with your own academic practices and choose your AI study tools wisely.