The excitement, The fallacy and The big steal. Thoughts on chatGPT
“Thoughts on creation and originality in an automating world"
With much excitement chatGPT has taken center stage. Children and adults alike have flocked to the command prompt of chatGPT and tried all kinds of queries. The imagination running wild, the possibilities seemingly endless and the horizons of what could be achieved with this boundless.
Observing the playfulness of my kids and their friends in applying chatGPT to solve questions of knowing. And then watching Yusuf Mehdi presenting the new features that bing unlocks in its latest ai powered iteration I’m struck with the realization of one big pink elephant.
The example of comparing the Q3 statements of GAPinc. and lululemon after summarizing the GAPinc quarterly statement is a showcase how this new machine can be a fantastic timesaver and at the same time it is the prime example of the major limitation of chatGPT.
After all the elation I ‘d like to focus on some unmentioned and underexposed problems to the excited news articles, lazy coders and SEO marketers. Yes, your SEO can now be dealt with easily. Coding moves from craft to knowing to ask the right question. I get it.
The fallacy.
In an abundance of training data faults in composing the correct answer to a given question is a matter of quantity derived accuracy. With the web containing Petabytes of data, abundance is all around us. However, mistakes or ‘hallucinations’ are a real occurrence. Work is in progress to incorporate ‘hallucination’ detection to catch those instances. Until then you better check the texts yourself.
The biggest fallacy however is believing that you can find new insights. I will put forward here my conjecture that the current iteration and paradigm in machine learning/ai are intrinsically bound to that which is already known and as such have no entropy, no information or surprise to offer. Users of the output of chatGPT and alike are utterly mistaken when they interpret something that reads new as being new.
The big steal.
The above conjecture is problematic in 3 profound ways.
1) We go to school, we learn, we read what is known and yet the capacity to generate new problem statements that in turn generate new previously unknown solutions, designs, or texts stem from our human faculty to be ‘amazed, scared, inspired, love’ and out of that create something new. Unsolicited. Not in a million years will chatGPT ever create. The command prompt will wait diligently forever. The chatGPT engine will never look with amazement upon its own writings and be inspired to start anew and create something out of its own vocation. It can’t be inspired. It won’t be inspired by its own hallucinations. (unfortunately?)
2) Allowing chatGPT and other data scavengers to disassociate the output or ‘composition’ from the origin is a clear case of plagiarism. Any output by these models is created out of preexisting ‘published’ knowledge. Good practice in all countries that have, and honor copyright laws is that mass plagiarism is not allowed. Having bing summarize a pdf for private use is not a problem. Having chatGPT construct a recipe or itinerary is clear plagiarism as it has originated from a source or reference that goes unreferenced. Where the old bing and google search results point towards the origin maintaining the link, this is now hidden.
3) The most profound problem however is the risk of output generated by these engines either get directly reintegrated into the data set or after being reconfigured by some human intervention. Where first Facebook timelines were recognized to be presenting self-referential adds strengthening the pull on the user into a stale and static rabbit hole of limited variety of which tiktok showcases this to the extreme. If SEO, information and business analysts etc. work gets augmented, sped up by machine labor this will lead to more content generated by a very limited number of re-hashing algorithms potentially turning a diverse environment into a static one. The big steal is that of originality and surprise.
The light.
Now, that all sounds very dark and some of it is quite disconcerting. Luckily, I’m also an optimist. Throughout history humans have shown that the propensity to scale brings forth the next step in mechanization, industrialization. Maximizing output and optimizing production that many time decimated the art of creating at each increment. The light, I have to offer is that each time, creators that understood how to use these new capabilities to their advantage built a new normal. There are great opportunities out there and I have begun to integrate these new engines into my own projects. Part of the human condition is that it is hard not to create.