For about the last week I have been obsessed with chat GPT. I literally have been exhausting my access to the application every day for about a week, testing out different approaches to asking questions of an artificial intelligence platform . And to me this is like another level of search where the responses somehow need to affect your cognitive understanding and fool you into thinking it’s a human response. So the level of communication is on one side very absolute and on the other somewhat dumbed-down as if to represent the mean response as opposed to a truly enlightened response. What you find is the results of your searches tend to generalize the information that the artificial intelligence presents. Objectively this Approach to communication is good for topics where an unbiased response is necessary but may be horribly inaccurate for communications with any cultural context such as music lyrics poetry and other forms of artistic or cultural expression linguistically. For the last week or so I’ve been testing various question and response themes including create a song, create a movie plot, Create a menu, create an itinerary, for example a seven day itinerary for a trip to Europe; and I have incorporated pop culture references of the full range to include the Wu-Tang Clan Rage Against The Machine Cancun Dwayne The Rock Johnson Chris Rock Rock Hudson and Jennifer Hudson and every time I am surprised by the results Chat GPT Returns. I have asked questions about what are the words women and men each want to hear the most, what would entail a perfect date, what the meaning of life is and even what is the key to happiness in life and most of the responses are something you would expect from a typical advice column with the main difference being the advice is from a robot and not a human. What does this promise for humanity in so much as well our interaction with artificial intelligence be the type where a black box surrounds the algorithms and the underpinning intelligence structure so we only see the results in the responses to the questions that we have for an artificial intelligence system or do we have the ability to affect the responses and the results that we derive from the artificial intelligence through a feedback loop where we can modify and alter the response based on our cognitive understanding of real cultural dynamic cues and social custom folkways and other forms of groupthink and cultural identity or is the point of artificial intelligence to desensitize the user of their cultural affect thereby neutralizing any cognitive bias created from cultural cues such as ethnic or social customs rooted in long-standing folkways and traditions? If we so derive results from an agnostic or so-called agnostic artificial intelligence platform and subject these results to the critical culturally affected nature of an authentic human being do we find that the inherent structural integrity of the artificial intelligence platform is by virtue of its meanness or averageness actually misrepresenting the populations of users who are relying on artificial intelligence to guide their decision making and if so is it a responsibility for the human agent or user and in this case recipient or consumer information from an artificial intelligence platform to ethically and responsibly guide the AI to consider those cultural affects as realistically as possible to the extent possible as represented in a population ?
I think the human responsibility is in the latter that we need to affect these artificial intelligence platforms by excepting and grading to the greatest extent possible as consumers of this information the artificial intelligence response in terms of quality data and structural integrity when compared to subject matter expert response from humans and the generalities knowing that a computer information system is going to favor heavily that in the general architecture of a thought as opposed to the more nuanced or culturally specific and stylistic variations of that generic same theme.
For example, I did a search for Chat GPT to create for me a Rage Against The Machine song about finger sandwiches and here was the “interaction”:
Hey ChatGPT…Write a Rage Against the Machine song about eating finger sandwiches.
Finger sandwiches, on my plate
Corporate greed, ain’t hard to hate
They try to keep us all in line
But I’ll fight back, with a sandwich in mind
Chorus:
Eating finger sandwiches, gonna break the chains
Fighting for my rights, with a sandwich remains
I won’t be silenced, or held down
Eating finger sandwiches, gonna wear the crown
Verse 2:
They try to control us, with their fancy foods
But I won’t be swayed, by their capitalist stews
I’ll take my finger sandwiches, and my cup of tea
And I’ll rise up, against their industry
Chorus:
Eating finger sandwiches, gonna break the chains
Fighting for my rights, with a sandwich remains
I won’t be silenced, or held down
Eating finger sandwiches, gonna wear the crown
Bridge:
I won’t be bought, by their expensive fare
I’ll stick to my sandwiches, and my fight will be there
I’ll take a stand, against their system of hate
Eating finger sandwiches, gonna seal their fate
Chorus:
Eating finger sandwiches, gonna break the chains
Fighting for my rights, with a sandwich remains
I won’t be silenced, or held down
Eating finger sandwiches, gonna wear the crown
Outro:
So next time you take a bite, of your finger sandwich
Remember to stand up, and make your stance.
Now how representative of a Rage Against The Machine song is this?
This we can test by asking a sample of individuals in a survey. We can ask qualitative questions about whether or not the lyrics in the ai response accurately represent a population. What is user sentiment of ai response to culturally derivative queries?
Not withstanding the above research question what does ai hold for the owners of culturally derived works where style is interpreted as influence and where does the line stand for title and claim to derivative work. If I ask ai to write a Stephen King type novel does Stephen King deserve (much less want) copyright? Moreso, if you ask Ai to create the perfect movie plot and you take that and produce a movie that flops, can you sue the ai platform for inaccurate response. Or does this mean the perfect movie is a flop? These are the sort of myriad questions ai may scare away users with. Then again, the idea of acquiring an easy prompt for an impromptu date or a 7 course menu in the french style without so much as a url worth of manual search? Oh yes! The fact is as ai gets smarter the responses will get better and better, but it will probably always have some discernable awkwardness. Is this ai generated disconnected cringe any more awkward than being human? Do we have realistic user expectations when engaging ai? Is there a certain tolerance or inherent bias in either the user or the ai?