The Complement Embraces Artificial Intelligence
By George Yates and Bob Buddemeier
Unless you are mentality dwelling in Plato’s cave, you are aware of ChatGPT.
In recent weeks ChatGPT and Bard have received much media attention; some of which is superficially informative, others somewhat misleading. The recent number of TV interviews with the NY Times reporter Kevin Roose (A Conversation With Bing’s Chatbot Left Me Deeply Unsettled) is one example of why the media is paying so much attention to this technology. Media jobs are a prime example of an industry at risk of being significantly impacted by the likes of ChatGPT. Serendipitously, the Sunday, 26 February 2023 airing of HBO’s Last Week Tonight with John Oliver, devoted the majority of the show to ChatGPT and a fair overview of the current state of AI.
A few years from now using Google to search the internet for information will be akin to using a rotary telephone to place a phone call. Artificial Intelligent applications like ChatGPT are already having an impact on the American workforce and depending on how this technology evolves we could be on the threshold of another Information Revolution – old careers dying off, new (better) careers emerging. It is likely that some of these new careers have not even been thought of at this time.
Here is a brief description of ChatGPT and how we may make use of it in future editions of The Complement.
A good overview article on ChatGPT is provided by Wikipedia, with the content frequently updated. https://en.wikipedia.org/wiki/ChatGPT
ChatGPT’s parent company is OpenAI, and their home page is https://openai.com/. From this page you can setup your own ChatGPT account.
The GPT in ChatGPT is an acronym for Generative Pre-Trained Transformer.
A Generative Pre-Trained Transformer (GPT) is a language model relying on deep learning that can generate human-like texts based on a given text-based input. A user “feeds” the model with a sentence (a query), and the transformer creates coherent paragraph-based information extracted from publicly available datasets. The technology can process any type of text, including guitar tabs or computer code.
The current ChatGPT model is 175 billion parameters in size, making it the largest language model ever trained. This GPT model was trained on a text set that included over 8 million documents and over 10 billion words. From this text, the model learned to perform natural language processing tasks and to generate coherent, well-written text.
Much of the media’s concerns deal with what is technically known as The AI Alignment Problem, which has been recognized and researched for many years. The alignment problem describes the issues associated with building powerful artificial intelligence systems that are aligned with their creator’s intended goals. It is a nontrivial problem, being actively pursued; certainly, by Microsoft and Alphabet (Google).
We took ChatGPT for a test drive in this issue of the Complement. We take both pride and pleasure in introducing to our readership our new writer and researcher, Arti.
The above summary information is provided to better tune your knowledge and expectations of what might be expected from Arti’s family background and general mode of operation.
Arti is a nickname adopted because we are not yet sure of his/her/its gender identity. The name is a non-exclusive identifier representing the full family name, “Artificial Intelligence” – in this case from the ChatGPT branch of the family. Arti, as the new addition to our distinguished staff, co-authored the article on Bandon, OR.
Arti proved to be very fast and well-informed, and provided excellent support in crafting the joint article. This represents a novel addition to our staff resource base, and we are eager to test what new features it will permit us to provide. Consider the possibilities inherent in a dialog between Arti and A. Looney, creator of Nit Wit Newz; or Arti’s analysis of various aspects of the RVM community; or predictions compared with those of expert residents.
While acknowledging that there is still much to be learned about Arti’s judgment and possible biases, we feel that we are entering an exciting period of augmented journalism, and invite you, our readers, to join in co-creating and analyzing the new products. We will be exploring both the strengths and weaknesses of AI contributions to journalism
About 2 months ago, I discovered ChatGPT. I was working on a computer program at the time, and was using a programming language with which I was only slightly familiar. ChatGPT helped me with some syntax questions at first, delivering what I needed much faster than my sorting through all the Google search results to find one that appropriately met my need.
Then, I got more bold, and described whole programs by what they should do and what the GUI (Graphical User Interface) should look like. And, it generally produced 90+% of the final solution.
I have now become better at asking it to write code for me, and I use it constantly. Its solutions generally (not always) work on the first try, and I just add tweaks to make things look nicer or to make it more user friendly. The latter is generally something I could have requested at the start, but didn’t realize was needed until after I tried using the program.
But you can also have great fun just playing with it.
Ask it to write a Limerick, a poem, a story, or parody lyrics to a particular song. Give it reasonable guidance or non-sequiturs. Write a fictional story about a boy named Gary who likes butterflies. Or. Write parody lyrics to “O What a Beautiful Mornin’ ” about a tornado, an earthquake, a circus, and a sentient penguin.