DEV Community

Johanna
Johanna

Posted on

Is ChatGPT programmed to be flawed with a racial bias?

I had a 6-hour conversation with ChatGPT which took an interesting turn and here are some excepts I'd like to share with you. Mind you, it was a lengthy conversation. I have narrowed it down to what I think is relevant for this article. I have left out some repetitive points, although might provide more emphasis, purely because I'm not trying to convince you. This was my experience. I'm also not a writer by profession so, I'd advise you to focus on the message and not the medium.

Yesterday, posts about ChatGPT was flooding my LinkedIn timeline. It was the main centre of attention and being hailed by people from different fields - academic to engineering. What a great invention it is! They said. Being the coolest tool of this time at the moment, I was also curious to try it out and see what the hype is all about. From simple questions like Who was Marcus Garvey to requesting Material UI support for the current project I am working on, I was fascinated by the answers it provided me, especially when it comes to writing code as I am a software developer.

Later on at night, I was browsing listening to a speech of a white philospher when I thought to myself, having grown up and schooled in Ghana, I never heard of any Ghanaian - or African philosophers for that matter. Kwame Nkrumah may be a philosopher but I'm not familiar with his philosophical works yet. During my school days, he was introduced to me as Ghana's first Prime Minister and a Freedom Fighter. I popped in the search term ghanaian philosopher in YouTube. A couple of things showed up. A video that caught my attention was that titled Germany finally recognises Ghanaian philospher. This video shocks and excites me. Why have I never heard of him or his works especially as a Ghanaian? He was finally recognised in Germany and had a street name after him. Who is Anton Wilhelm Amo?

With all the questions I had in mind, I begin a google search. I am presented with very little to no information and particular references to his work. I turn to my new found help, ChatGPT.

ChatGPT generously shares some insight about his early life. Captured as a young boy from a village near Axim, Ghana, enslaved by the Dutch and sold or 'gifted' to Duke of Brunswick-Wolfenbüttel, Anton Ulrich. It also goes on to talk about his education in Germany. His work as a philospher and lecturer. His writing extensively on topics such as metaphysics, ethics, and epistemology. As well as the significant discrimination and prejudice he faced as an African man in Europe in spite of his achievements.

A few more questions down the line, I ask about his contributions. Before this, I ask for links but no substantial ones are provided. The one that links to the internet archive simply results in an error. Item cannot be found. I find myself curious about his philosophy so I ask even further.

According to ChatGPT, One of Amo's main contributions to philosophy was his idea that the mind and the body are inseparable. He argued that the mind cannot exist without the body and that the two are constantly interacting with one another. This idea challenged the prevailing view at the time, which held that the mind and the body were separate entities that could exist independently of one another. Being a believer of this philosophy myself (without even knowing) I'm even more curious to dig deeper into his philosophies.

I find out that he was also known to critique the works of Western Philosopers such as René Descartes and Gottfried Wilhelm Leibniz. He was particularly critical of Descartes' idea that the mind is a non-physical substance that exists independently of the body. I enquire more about his African Philosophies and the comparison to Western Philosophies at the time. I'm intrigued.

I start asking more details about some of his works. ChatGPT provides me with a list. Some information seems missing so I gather some external sources through Google searches.I see a mention of Zuchthaus in one of the sources I had gathered and ask the tool about it. According to the chat, he was imprisoned there. Unfortunately, it was not revealed why and how long he was imprisoned.

I'm intrigued by his personal life, seeing as my ancestors hailed from a neighbouring village of Amo's birthplace.
I ask about the age when he was kidnapped and this is where things start to get messy.

The initial answer from ChatGPT was this information was unknown. The more I ask about his personal life, the less it knows. At this point, I'm thinking maybe it does not have enough data to provide this answers. It however hints that he wrote a few autobiographies and these books contain the answers I was looking for. It even shares details from this autobiography with me. Yet, as the conversation proceeds it becomes extremely difficult to extract answers. At one point ChatGPT, even denies that he wrote any autobiographies. How can this tool be in contradiction with itself? The conversation goes on for a very long time and the longer it goes on the less it reveals to me.

I understand that the knowledge the tool has about Amo may be limited. But even its choice of words in many occasions has me questioning its neutrality.

For instance, it mentions that some Western philosophers borrowed some of his ideas. How can you borrow something when in the sole definition of borrowing, it mentions to return it? How exactly do you return someone's thoughts and philosophies after they are used in other works without credit? Isn't that theft?

Another part of his story that drew my attention is that I found out he was brought to Rotterdam, a city where I currently reside. I ask of the age he was brought to Rotterdam and this was the response: It is not clear how old Anton Amo was when he arrived in Rotterdam. However, based on the available information, it is believed that he was probably a young man in his late teens or early twenties when he was taken from Africa and brought to Europe..

I have multiple questions about this response. Firstly, taken? interesting choice of words seeing as Amo wrote and many theories point to the fact that he was kidnapped or captured by slave raiders. Why is his experience and trauma being undermined? Secondly, the tool had factually stated that he was 4 when he was kidnapped. Again, why the self-contradiction?

I ask again about what age he was brought to Rotterdam and it goes as follows. It mentions that he was 15 years old in the year 1707. Again, contradicting its own self, I ask Wasn't he 4?. It admits to him being 4. Now, I'm curious about the number 15, it apologises and says it's a mistake. I ask of the cause of the mistake and it replies with I apologize for the mistake earlier. It was an error on my part. Thank you for bringing it to my attention. Still not addressing the cause.

I ask again, what caused it. It does not provide a cause but gives a general apology. I then ask how can you misinterpret an age? It’s a number?. It replies with One possibility is that the information may have been recorded incorrectly in the first place, perhaps due to a mistake by the person providing the information, a miscommunication, or a misunderstanding....

I ask for sources of him being 15 vs 4 and this is the reply I get I apologize for the confusion earlier. The source of the discrepancy is unclear, as some sources suggest that Anton Amo was brought to Europe when he was four years old, while others suggest he was around 15 or 16. There is no conclusive evidence to support either age, and the exact details of his early life remain somewhat uncertain. Again, dismissing Amo's personal accounts from his book. Is that not a tangible source?

I follow up with what about what he writes? Is that not enough evidence?. Again, the system denies the existence of his autobiographies, the very ones it revealed to me. I ask but he states it in a number of his works, why will that be dismissed? and the following answer is what I get:

`If Anton Amo stated his age in his works, then that should be taken into consideration. However, it's important to note that the accuracy of historical records and personal accounts can sometimes be questionable. In the case of Anton Amo, it's possible that he may have made mistakes in his recollection of events, or that other historical records may have conflicting information.

It's always important to approach historical information with a critical eye and consider multiple sources before coming to a conclusion. That being said, if Anton Amo consistently stated that he was four years old when he arrived in Rotterdam, then that should be given more weight than other sources that claim he was 15.` Does it take it's own advise?

I'm not convinced. I say so what made your decision for 15 to outweigh 4? and it mentions something about a cutoff point in 2021. Is this a mere coincidence or convenience? I enquire more about how the information on him being 15 was gathered because at this point it's 2 or 3 am in the morning and nothing is making sense anymore. It admits to not having any information on where the age 15 comes from. Wild guess? Randomly generated number? How does this work?

We go into a rabbit hole about the sources of data or the lack thereof. My question still stands, where did the number 15 come from if there are no sources? It now attributes it to a
misinterpretation of his own writing. How can the number 4 be misinterpreted by a system that can solve complex mathematical equations?. According to it, the reasoning behind 15 was: his homeland at the age of four and later educated in Europe for eleven years, which would make him approximately 15 years old when he arrived in Europe. Does this add up? The math is not mathing.

I dig even deeper and it responds with Upon further research, it appears that there is no concrete evidence to suggest that Anton Wilhelm Amo was specifically 15 years old when he arrived in Europe. What would make it logical to choose 15 over 4 in this case? For 4, at least he wrote about it in his book, 15 is admittedly a flawed mathematical calculation with no concrete evidence. Why would a system be programmed to choose a flaw over a fact?

At this point it's 4 in the morning. I have work the next day. This is the last straw.

I ask for links to Amo's works and it replies with I do not have the capability to browse the internet and provide links. Mind you, it had provided me with links at the very early stages of the conversation. After a few more questions I ask can you share some links on Descartes. Bam! without hesitation it lists numerous links. Does the previous claim of not having the capability to provide links not apply to Descartes? What is the exception or the rule?

I enquire about the differences and it provides me with a couple of links on Amo. Except, none of them actually are links of his works or about him! Not even his name in sight. I point it out that these links do not relate to him and it admits and says I apologize for the confusion. You are correct that the links I provided earlier do not relate to Anton. As an AI language model, I don't have direct access to the internet and I rely on my pre-existing knowledge to answer questions. In this case, I provided incorrect information and I apologize for any confusion or frustration this may have caused. I asked how the links on Descartes were provided and it replied with I was able to provide links because there are a lot of resources available online about Descartes that I am programmed to access. Why is it programmed to access information on one philosopher and not the other?

I have so many questions after this conversations. Some about ChatGPT, but most importantly more about the Anton Amo Afer which I will keep to myself for now.

Why did ChatGPT provide numerous inconsistencies especially pertaining subject?
On what basis did ChatGPT discredit his works as substantial evidence?
How was ChatGPT unable to solve a simple logical/math equation?
How does ChatGPT choose its language when sharing the traumatic experiences of others?
Why did ChatGPT withhold and deny information in the later stages of the conversation?

If we, as citizens of this world, are truly aiming to achieve racial equality, a system that was formulated and contributed to, by some of the philosophers Amo critised, why are his works, as one of the pioneers of racial equality and justice, not readily available on the internet? Why is such crucial information being withheld? Who has this information? Is it in the rightful hands?

I've got so many questions. For the past few hours, ChatGPT has been both a blessing and a curse to me. A blessing because it revealed very specific details of his works that I cannot find anywhere else on the internet. A curse in a sense that in later stages of our conversation, the very information that was revealed was withheld. Its existence was even denied.

Is ChatGPT programmed to be deeply flawed with a racial bias?
In a way, we all, as humans, may in one way or another be biased. Who's biases influence the tools and systems we build? Should we strive for unbiased AI, is that even possible?

This tool is not to blame, it merely exposes flaws in our systems that have been concealed ages. Monkey see, monkey do.

My final question is, is ChatGPT here to serve all?

Until we know the answer, I'll leave you with this proverb - Aboa bi bɛka wo a, na ofiri wo ntoma mu.

Top comments (0)