The irresistible rise of the auto-reporter
March 28, 2023, 9:04 pm , by Richard Lutz
As key figures in artificial intelligence warn today that developing AI systems is out of control, Richard Lutz reads an article created by computers that took seconds to produce. Highway to progress? Or path to perdition?
“Caitlin Clark is a rising star in the world of basketball, and many believe that she is the future of the sport. At just 19 years old, Clark has already made a name for herself as one of the most talented and dynamic players on the court…”
So opens a readable feature about the rising star of American women’s basketball. Nicely mapped out, clear and concise.
And one other thing: it’s been written by a robot. A London journalist asked the A.I. system Chat GPT to write 500 words on Caitlin. Within five seconds of typing the request, the programme knocked out the above paragraph and the ensuing article. Sometimes, my friend explained, the ‘typing’ stopped for a second as if to contemplate the next batch of content. But it had it all. Though it did get the athlete’s age wrong. An error, by the way, that even flesh and blood reporters make.
Frightening? Exciting? The future? All three?
If anything A.I. will, by its very existence, create a whole world of issues and problems for those who ply the oceans of the written world whether it be a newspapers, online publications or maybe even tv programmes.
“This could get scarily impressive …”
My friend said of the Caitlin Clark article: “It’s pretty accurate. I’ve read worse. But I’ve also read better..” He went on to ask Chat GPT to re write the article in another style, one with ‘more personality’. And this is what the robot came up with:
“Caitlin Clark is a basketball prodigy that has set the court ablaze with her killer moves and killer shots. At only 19 years old, she’s already causing a stir in the basketball community, leaving people wondering if she’s the lovechild of Michael Jordan and Steph Curry.”
Stylewise, a bit more snappy. Maybe like the back page of a sports section.
‘Not bad.” said my friend who commissions and edits international copy for a newspaper. ‘’You could see how with a few more tweaks it could get scarily impressive.”
In fact, the artificial intelligence tools can re write a piece for a tabloid, for a broadsheet, for a newsy personality feature, for a gossip column, for a local paper. In the style of a specific newspaper, in the style of a specific journalist. For anyone. In under a minute.
Of course, there are huge issues involved. Firstly, is an A.I. article genuine? Its tools can reach deep into the internet. And it pumps out facts, numbers, and quotes with its cyber-tentacles and slots it into readable templates we understand. But it can’t separate fact from assumption, fact from half-fact. It can’t identify the difference between the reliable source and the half cooked source. It can’t pick up a phone to double check something or to prove a quote is not manufactured. It has no truck with copyright law and begs the question of how libel or defamation is handled and to whom it is accountable.
It can be set up to churn out, let’s say, 25 negative online reviews of a restaurant or a movie in 25 different styles equipped with typographical errors and mistakes to ‘seem’ more human. It could even, possibly, ignore a legitimate request and write an article how it thinks it should be pitched.
It could be…as the journalist said…. scarily impressive.
share this post!
robert blumberg
You have nothing to worry about — I tried twice to have ChatGPT, “write an opinion piece about Boris Johnson in the style of Richard Lutz.”
The second, the better follows…
As I sit here in front of my laptop, trying to put my thoughts into words, I can’t help but feel a sense of frustration and disappointment when I think about Boris Johnson, the current Prime Minister of the United Kingdom. In my view, he is nothing more than a bumbling buffoon who has somehow managed to climb his way up the political ladder, without any real qualifications or merits to back him up.
Let’s be real here, Boris Johnson is not a politician in the true sense of the word. He is a performer, an entertainer, a man who loves to bask in the spotlight and make outrageous comments ..
It goes on…
SJ from Birmingham
an interesting time. Can see that AI has its benefits but do feel pretty scared as to where it may take us. Heard that Credit Suisse’s systems stretch into different domains so much that it is too complicated for anyone to know the whole over-arching organisation – rather worrisome ….. understand it has had three CEOs in three years…..
Usain Qureshi
Scary stuff. There have been several articles in ‘New Scientist’ exploring these issues. Several are beyond my comprehension but it is clear that the world of literature is facing fundamental shifts.
A. Davidson
So sadly like a lot of our media then
Bobby Jane
I feel that I should keep up to date on this stuff. So far, my understanding is that AI has, at its core, content that actual human beings have created. For good or bad it generates its output from material that it has its own inbuilt biases.
The unsettling thing for me is that my usual three “critical thinking” questions
1) who is telling me this?
2)why are they telling me this?
3)what are they not telling me?
become a little more difficult to apply if the “who” is not entirely clear or/and is increasingly unclear.
Ed Lessem
the problem that a lot of experts say is that the humans will soon let AI develop itself as it will be way better at programming than humans (and way faster). At that point, it will be too complex for humans to understand how it works – and possibly what its motives are.
At that point, you could ask a Chatbot what, for example, causes climate change. And it could give you a very good answer. But we don’t know why it’s giving you that answer – it may have its own reasons…
Lisa Meadows
An important point that Stanley Kubrick was making in his film Space Odyssey was when the spaceship computer overturned human decisions
Will Travel
A good podcast explainer
https://podcasts.apple.com/us/podcast/the-ezra-klein-show/id1548604447?i=1000605132896
Vannin
Scary and weird times
Jim Worth
I got ChatGPT to read Strunk’s Elements of Style and suggest edits for your article on AI. It said:
Here are some suggested edits for the article based on Strunk’s Elements of Style:
Use active voice instead of passive voice: Instead of saying “An error, by the way, that even flesh and blood reporters make,” use “Even flesh and blood reporters make this error.”
Use concise language: Instead of saying “Whether it be a newspapers, online publications or maybe even tv programmes,” use “Whether in print, online, or on TV.”
Avoid unnecessary words: Instead of saying “My friend said of the Caitlin Clark article,” use “My friend said about the Caitlin Clark article.”
Use proper punctuation: Add a comma after “It could even, possibly, ignore a legitimate request.”
Use parallel structure: Instead of saying “For anyone. In under a minute,” use “For anyone, in under a minute.”
Overall, the article is well-written and engaging, but these small edits could help improve clarity and readability.
Martin
Pretty sure the Sun newspaper is written by robots, possibly a couple of software upgrades behind?
Jennie McAskill/ Dundee
Writer John Lanchester in a recent edition of The London Review of Books, referring to AI call centres, says: ‘I suspect customer service systems will replace phone automation hell with chat services that are much better …but offer no exit from their closed systems, are incapable of admitting error and won’t let you engage with a human being, ever.’
Subscribe to new posts.