If youre like me, you have your favorite authors and cant wait to read their new releases.
You know what to expect based on their original style and characters.
Likewise, we gravitate toward a favorite poet, playwright, scriptwriter and lyricist.
But, what if you dont know who wrote the book? The poem? The play? The movie? The song?
What if it wasnt created by a human at all? But by an artificial intelligence tapping away on the keyboard and spewing out in minutes what it might take a person hours, days, weeks, or years to produce?
This isnt some futuristic sci-fi concept.
Its real.
And its happening right now.
OpenAI, a San Francisco-based company, launched ChatGPT last November.
GPT stands for generated pre-trained transformer. Using an LLM, or large language model, it can draw on countless databases to produce a coherent story, essay, or other artform.
Its capable of creating new works of literature. Imitating someone elses style like Shakespeare, Dickinson and Whitman. Or cranking out its own unique work to comply with any theme thats requested.
All someone has to do is give the AI an assignment. Set some rules. and in mere minutes it can spit out material on demand.
It can crank out a sonnet.
A haiku.
A scene.
All by analyzing and mimicking human writing and simply stringing together words, phrases, and sentences that seem to follow a logical progression.
The AI doesnt understand what its written. These are just random words pieced together.
Maybe it makes sense.
Maybe its nonsense.
But it probably wont replace human writers.
At least any time soon.
Actor Alan Alda decided to put an AI to the test.
Alda played Hawkeye Pierce, the smart-aleck surgeon serving during the Korean War in the hit TV series, MASH, which aired from 1972 to 1983. He asked the AI to write a humorous scene for him and his series buddy, B.J. Hunnicutt, played by Mike Farrell.
Alda instructed the AI to focus on an argument between the two surgeons where Pierce accuses fellow prankster Hunnicutt of stealing his lucky boxers.
In a March New York Times article by Julia Jacobs, Alda said he asked for snappy humor and lively dialogue. Alda complained that although he requested funny, he actually received really stupid stuff. Even sappy and sentimental where the AI goes off book and says something silly about Pierces grandmother. It has a terrible sense of humor, Alda concluded.
A Washington Post article by Seth Perlow agrees that AIs dont pick up on jokes and puns and their writing has been criticized as being soulless, barren, cliched, and predictable.
Scott Simon, host of NPRs Weekend Edition aired a show last December that focused on machine-made poetry. Poet/critic Katha Pollitt studied some AIs writing and concluded it was cliched and tiresome.
She added that this was to be expected since AIs do not have experience, emotion or imagination.
Calum Chase, an expert on artificial intelligence, told Simon that the appeal of AIs is they can write fast and dont have problems that human writers face, like writers block, anxiety, fatigue and distractions.
Chase says AIs may not be human, but their software is evolving.
So, who knows what the future has in store.
For now, one million people have reportedly tried ChatGPT, which is advertised as a free service. I started to sign up for one of the free accounts, but stopped when I was asked to submit a credit card. They swore it was for legal reasons and promised they would never charge anything to my account.
Still, I chose not to continue with a personal experiment.
As with many technological advances, there is the potential for abuse with AI writing. A major concern is how easy it is for students to cheat on various assignments. They could ask the AI to write their term papers. Or their college application essays.
Who would know, since admittedly it would be difficult to detect.
Even harder to prove.
And once something like this starts, its tough to stop. Especially since the service is so easy to access.
The question is who should regulate this?
Who should be the watchdog?
Its really not the governments business. and legally its not breaking any current laws unless the process stumbles into the plagiarizing zone.
Theres also little incentive for the industry to regulate itself. Or establish an ethical code of behavior.
One proposed solution is to set up an independent review board to police the new technology. and to collect text generated by commercially available LLM and place it in an independent repository, making it easier to detect plagiarism.
Another precaution mentioned is to put age restrictions on who can access the software. That forces students to write their own essays and term papers and not rely on AI to do the work for them.
Interestingly, instead of considering AIs as competition, some writers have actually partnered with them on projects.
Who knows, the AIs might even develop a funny bone based on those collaborations.
Read more:
Artificial intelligence | Strictly Opinion | richmondregister.com - Richmond Register