Think Twice Before You Use ChatGPT: The Hidden Dark Side of AI

Photography by Tara Winstead from Pexels

(TLDR at the bottom of this article. Thanks, algorithms.)

Yeah, yeah, yeah, I’ve heard about how amazing AI is and how quickly it can answer questions and write essays. But just because I’m a downer, no, because I value critical thought and hard work, I can’t help but feel like this will make us more reliant on technology to think for us, making us weaker thinkers when our greatest advantage as human beings is our brains. In the past half year, there has been incredible buzz around chatGPT and other AI. Sure, they can do a great deal, but have you considered the drawbacks that few are talking about?

Here are some concerns you should consider regarding AI.

Hidden Costs

There is always a (hidden) cost. No matter how much tech companies laud the benefits of AI and how this is a new era (I don’t doubt that), there are good things that are being undermined and devalued. The promises of AI are too good to be true. As I learned in Econ 101, there is always an opportunity cost. And I think the opportunity cost this time is the value of hard work, discipline, learning from firsthand, legitimate sources, creativity, thinking and intelligence.

Taking Shortcuts

AI is disruptive. I have no doubt about that. But I can’t help but think how some will use chatGPT as a shortcut by asking it or any of its variants to do the work of research and writing for them. Teachers have to find ways of evaluating whether students’ work is actually done by the student or by AI. GPTZero is one such tool, but even AI detection tools can be wrong.

People will no longer have the training and proficiency to do their own research, instead relying on what a machine that doesn’t know what it’s saying. And experts in the field can clearly spot errors and flaws in AI generated articles meant to inform its readers.

AI Hallucination

It’s a text prediction model, meaning it is simply predicting what the next best word is and sometimes that results in utter nonsense, jargon or generic writing that looks like every other AI-generated text. Even with specific prompts, it still lacks the depth of human perception. Oftentimes, the information AI presents is simply untrue. They call this “hallucinating.” Now I don’t know about you, but I would not go to a person who “hallucinates” for accurate information. Would you?

Concerns Regarding Accuracy and Bias

Additionally, you may be hard pressed to check its facts since AI doesn’t cite its sources. It could be getting information from anywhere on the internet prior to 2022. Even if you prompted it to cite its sources, you would still have to go through each one to make sure the sources aren’t misinformed or biased.

Ethical Issues

Furthermore, AI is not sustainable. Sure, businesses can get more work for less money, but hear this: Open AI employs low-wage workers. One article states that Kenyans were paid $2 hourly to make chatGPT safe for users while psychologically scarring the workers themselves. Another article discusses the large water footprint AI requires to work.

There are also questions about what one should do regarding generative AI that can clone one’s voice and image. This is an important issue for actors, voice over artists, models and the like to think about before gaining employment. As for companies and agencies, should you use employees’ image and voice even after they have left the company? For how long and for what reasons? What sort of legal documentation would this require? These questions are still up for debate.

AI Training

AI requires creative input. If it continues to rehash old information it will eat its own tail, my friend in the tech sector told me. Without the work of past billions of humans, AI is useless. It still requires humans to train it, and a good portion of the data used to train AI and make a profit was taken and used without its owners’ consent. Such was the case for DeviantART’s AI art generator, DreamUp, received backlash for that particular reason, although now they have taken steps to protect artists’ intellectual property through “noai” which allows creators to opt out of their work being used for AI training purposes. For what it’s worth, I’m not sure I want to feed this beast. We cannot guarantee that confidential information or your intellectual property will not be abused if you input it into chatGPT or other AI.

A startling report found that the quality of answers chatGPT-4 gives fell drastically from March to June of 2023. It’s accuracy in identifying prime numbers fell from over 95% to just above a paltry 2% over this time period. Its performance also fell in code generation, answering sensitive questions and solving math problems. The question remains as to why this is the case. Mark Tyson, the author of this article, also suggests that OpenAI monitors and publishes the quality for its paying customers. Not a bad idea. I might even go as far as saying that OpenAI should publish these quality checks for the general populace to see as GPT is being used across sectors by people from young students to businesses and plays a significant role in today’s society.

Finally in regards to training, the authors of materials AI is trained on did not give consent to allow their material to be used to teach AI. Two award-winning authors are now suing OpenAI over this issue. While work that is public domain can understandably be used to train AI, questions remains as to the legality of using copyrighted works for AI training.

This is especially important when it comes to artwork, with some artists even being accused of using AI to create their pieces. Ben Moran was banned from Reddit for this exact reason.

Fast Art Cheapens Artistic Creation

AI like Midjourney, with their quick turnaround and low costs, is decreasing the revenue artists make simply because they cannot compete in regards to cost. With cheap, fast and custom AI art, those who don’t value man-made art will turn to AI. AI may disincentivize human creativity and creation, one tech and art enthusiast says.

Additionally, AI generated content cannot be copyrighted. The U.S. Copyright Office has a document on AI policy which states work that is exclusively done by AI cannot be copyrighted. Only works that have “creative contribution from a human actor” can be registered under copyright laws.

Capabilities and Shortcomings

AI is just a machine. It can be a helpful tool, and I think students and the general public should be educated in its strengths and weaknesses so we can best decide how to use it. But AI does not and cannot understand the human experience. It does not have or experience emotions, have memories, or know what it means to be alive. AI cannot replace you.

Is that important? That depends. I wouldn’t go to ChatGPT the way I might go to a friend or professionally licensed counselor (or even a pastor — yikes!). AI cannot replace human relationships, as movies like Her and Ex-Machina explore. AI cannot care about us as a human being can. It cannot empathize or feel sympathy for you or me. This is important as interpersonal relationships are the number one indicator of longevity. Yes, you heard that right. Relationships play a significant role in our health and lifespan (Yang et al., 2016). I’m sure you’ve felt the physical effects of being around someone you don’t like or trust compared to how well you feel around good friends.

What Has Been Unleashed?

AI is reminiscent of Pandora’s Box or the Tree of Knowledge in the Garden of Eden. I fear we have unleashed something with unknown and unpredictable results upon humankind and I am not alone in thinking so. We no longer have to imagine what the future might be like because we are living a futuristic sci-fi novel.

I find myself agreeing with Big Tech for once, as well as the “godfather of AI,” Geoffrey Hinton, as they call for a temporary halt to AI advancement beyond GPT-4 until we can figure out what this means for humanity and jobs, and how it ought to be legislated. We are not prepared as a society nor as individuals for such an escalation of technology and the challenges it presents. Is it really a good idea to put this in the hands of minors or those may have less than savory motives? We should take a moment, breathe and reflect on what we have learned thus far and consider how much more power tech labs should bestow artificial intelligence. Just as you would carefully plan and use caution when venturing into uncharted territory, we should consider the risks and dangers of the uncharted plains of AI technology on an already vulnerable population.

As much as AI has helped us, it can also be used for terrible things. As the saying goes, with great power comes great responsibility. There have already been reports of what I am going to call voicefake AI being used to create deceptive hostage recordings in spite of the supposedly kidnapped girl not having a social media presence, not that this is the only technology that has been misused to terrible detriment.

I’ll end with a quote from Molly Weasley that has been repeating in my head like a broken record: “Never trust anything that can think for itself if you can’t see where it keeps its brain.” Thanks, Rowling.

TL;DR

I have many reasons for cautioning the further, unrestrained development of AI which has exploded in the past several months.

1. AI devalues hard work, learning, critical thinking and proper research.
2. Students will cheat and countermeasures remain flawed.
3. AI doesn’t understand what it’s saying.
4. AI is often confidently wrong (it “hallucinates”).
5. There are concerns about the ethics of those employing workers to work on AI.
6. AI’s sustainability is questionable.
7. AI writing is jargony, generic and lacks depth.
8. AI cannot replace experts who are experienced in their particular field.
9. AI does not cite its sources.
10. AI sources may be biased or misinformed.
11. AI requires creative input and harms artists’ paycheck.
12. AI generated work without human contribution cannot be copyrighted.
13. AI fundamentally does not understand the human experience. It lacks emotions and memories among other things that make us human.
14. AI is not a replacement for human relationships, professional therapy or pastoral counseling.
15. AI is uncharted territory and its long term effects are unknown. We are not prepared for increasingly advanced AI.
16. AI gives flawed human beings too much power and is already being abused by malevolent actors.
17. The inner workings of AI are unseen. It is, in effect, a black box.

Thank you for reading and I hope this article sparks critical thought and discussion on our use of artificial intelligence and its role in our lives.

If I have been misinformed or if any information is outdated, please let me know in a comment. Thank you.

Updated June 22, 2023: added voicefake AI
Updated June 29, 2023: added copyright guidelines
Updated July 7, 2023: added to ethics rights regarding generative AI
Updated July 9, 2023: added more about copyrighted work being used without consent to train AI
Updated September 17, 2023: added DreamUp and noai, video on AI art generators
Updated September 28, 2023: added decline in response quality

That was pretty heavy… here’s something lighter that will make you laugh!