Scribbling in the Margins - things that ChatGPT can't do
Action movies from the 1980's were awesome... maybe not all of them, but certainly the clutch of sci-fi action movies contained at their core Alien, Predator and Terminator. In each, and apparently unstoppable killer from another place or time is ultimately defeated by plucky humans who work out some kind of blind-spot/design flaw/trick that can be played against the seemingly undefeatable foe... normally after it's got them cornered, and has wiped out the rest of the team.
As an author and researcher, ChatGPT could be seen to represent the same kind of existential threat to our skillset as a Predator did to Schwarzenegger's special forces soldiers... so what's the equivalent to smearing ourselves in mud at the side of the ravine in a back to basics duel to the death? Well, there are a few things that I've tried recently that I'm sure a machine would struggle to match... the common theme to both is emotional intelligence and first-hand empiricism.
Eulogies
I'm not talking about obituaries (ChatGPT does a reasonable job of telling you about the life and works of notable figures from the past few hundred years... just ask it to tell you about Henry VIII in the style of a gossip columnist if you don't believe me), but rather the short, pithy summary of a person read out at a funeral - the ones that need to give maximum impact in minimal time, sometimes for a character who wasn't universally loved (again, a machine might have got away with Queen Elizabeth II, but Prince Andrew is a whole other ball game) are a uniquely human form of expression at the moment.
ChatGPT is a long way from making emotionally aware value judgements and being able to communicate them idiosyncratically in a way that builds empathy, which takes us nicely to our next niche.
Comedy
ChatGPT can tell you a joke, but what it can't do is draw on any real world experience or insights to make the telling of a tale laugh out loud funny. Rich, metaphor and comic timing are again, currently beyond the platform's capabilities and are likely to be for some time. The rise of an automated sense of humour is a long way off, with the platform currently sat at the scene in Robin William's on- screen performance of Azimov's bicentennial man where the robot is asked to tell a joke. Comedians and satirists are probably the safest of all writers at the moment.
Investigative reports
Anything that requires the writer to go somewhere, experience something and write about what they've seen is (currently) beyond the scope of ChatGPT and similar platforms. I've been out and about on the ground performing cyber-physical inspections and pen-tests. It's fair to say that until someone gives ChatGPT a body that part of my work is safe as well.
As for the rest? Well, Prakhar Mehrotra gives one of the best summaries of the impacts chatGPT and similar technologies are going to have on writing as a profession, discipline and way of life. I certainly wouldn't want to be trying to make a living off articles like "5 must-see natural wonders in the UK" right about now!
When someone does give ChatGPT a body, I'll do a follow up article on how to beat it in a fight that may involve mud, ravines and an inimitable perchance for pithy one liners at the moment of dispatch... because, fundamentally, what ChatGPT will always be missing is the ability to learn from it's own experiences and extrapolate based on a uniquely personable real world model and (I hope) a capacity to do violence on the cyber-physical plane.
Enabling others to reimagine how to lead selflessly, reinvent solutions, and repattern socially desirable outcomes
1 年So true Andrew, and I also worry for the integrity of academic honesty (not artificially produced plagiarism!)
Cybrarian | ND & Pride Ally?? | Humanist | BeyondTrust.com
1 年A great article once again Andrew and takes me right back to Searle's #ChineseRoom