Can Bloggers Be Replaced by AI?
You be the judge.
The big news in national security circles the last 24 hours has been the report of a Chinese spy balloon discovered hovering over Montana. Having no real insights into the matter, I decided to give ChatGPT a chance to provide 800 words:
In recent news, a suspected Chinese spy balloon was found floating over northern United States. The discovery of this suspicious object has sparked concerns about national security and has led to investigation by the Federal Bureau of Investigation (FBI).
According to ABC News, the object was first spotted by a group of hunters in North Dakota. The hunters reported the sighting to the local authorities, who then notified the FBI. The FBI has confirmed that the object is a type of high-altitude balloon, but the agency has not yet determined the origin or the purpose of the balloon.
The FBI has stated that they are working with the Department of Defense to determine the nature of the balloon. They have also issued a statement, saying, “At this time, there is no evidence to suggest that the balloon poses any threat to national security.”
However, this statement has not stopped speculation about the balloon’s origins and purpose. Many news outlets are speculating that the balloon may have been sent by the Chinese government for the purpose of gathering intelligence.
This speculation is based on the fact that high-altitude balloons are commonly used for reconnaissance and surveillance. The balloons are able to reach altitudes of up to 20 miles, which allows them to gather intelligence from a large area. This technology is not new, as countries have been using these balloons for intelligence gathering purposes for decades.
In addition, the fact that the balloon was found over northern United States has led many to believe that the Chinese government may be targeting the United States for intelligence gathering. The northern United States is home to many military installations, including the Minot Air Force Base, which is known to house nuclear weapons.
The discovery of the balloon has also raised concerns about China’s increasing military capabilities. The Chinese government has been investing heavily in military technology, including high-altitude balloons. This has led many to speculate that the balloon found over northern United States may be a part of China’s larger military plan.
Despite the speculation, the FBI has stated that they are still investigating the balloon and have not yet determined its origin. The agency has also stated that they will continue to work with the Department of Defense to ensure the safety and security of the United States.
In conclusion, the discovery of the suspected Chinese spy balloon over northern United States has sparked concerns about national security. While the FBI has stated that the balloon does not pose a threat, the agency is still investigating its origin and purpose. The speculation about the balloon’s origins and purpose is based on the fact that high-altitude balloons are commonly used for reconnaissance and surveillance, and the discovery of the balloon has raised concerns about China’s increasing military capabilities. Until the FBI determines the origin of the balloon, speculation and concerns about national security will continue.
While not the worst summary I’ve ever seen, I am not impressed.
Well, it’s more literate than most of the human commenters at Lucianne.com.
Does Captain Obvious publish a blog?
Meets or exceeds what we get from mainstream and social media these days. Whether old-school bloggers are threatened isn’t really indicated. And there are so few old-school bloggers anymore.
@JKB:
So, not actual reporting at all, but a rehash of MSM reporting is, in your opinion, superior to the MSM. Which ChatGPT merely summarizes.
Will it ever occur to you that you’ve been brainwashed to distrust the MSM so that you can be more easily programmed by Fox, Alex Jones and your favorite Twitter racists? Hmm? No? Because at the point where you think an AI summary of the MSM is somehow superior to the MSM, you’re either a cretin, or a victim of brainwashing.
Written better than a lot of reports I had to grade from HS students. 😛
Let me know when ChatGPT can come up with conjoined twins using nanobots to rewire brains and create a worshipful cult trapped in the empty spherical tanks of an LPG ship.
AI can only use the rearview mirror, it is not (yet) capable of originality.
@Michael Reynolds:
Now I’m wondering how ChatGPT would respond to “write a short story about conjoined twins using nanobots to rewire brains and create a worshipful cult trapped in the empty spherical tanks of an LPG ship”
@Stormy Dragon:
Wonder no more:
I’m sure commenters, except at OTB, could easily be replaced by chatbots, but for conservative sites you’d have to have AI with no access to anything but FOX and points rightward. Bloggers, on the other hand, maybe they could be replaced, maybe not. But how would you ever know?
Isn’t AI supposed to be good with grammar? We say “over northern Canada” but “over the northern United States”. And does the FBI really have responsibility for balloons at 50,000 feet. Who they gonna fail to interview?
NPR reports on asking ChatGPT to do some simple rocket science. An area where results are verifiable. It should have stuck to nuclear psychiatry.
And this is not new. The AI noted, I believe correctly, that recon balloons have been routinely used, but also in WWII the Japanese attacked the continental U.S. with, I kid you not, Fu-Go balloons.
I’ve seen worse…
What is an LGP ship, and why did it carry so many conjoined twins and brain nanobot experts? 🙂
@Kathy:
https://en.wikipedia.org/wiki/Gas_carrier
“You will be upgraded”
I’ve seen worse student essays, TBF.
@JohnSF:
Oh, yeah. Far worse.
@CSK: More literate than Lucianne isn’t much of a bar to jump, tho.
@Mu Yixiao: Again, not a particularly high bar. I’ve had HS students who couldn’t have written that summary without pretty extensive instruction.
.
@CSK: Then again, I didn’t give my students a prompt with 5 or 6 elements in it. Give ChatGPT just the conjoined twins and the LPG tank and see what it comes up with.
ChatGPT 1) can only reproduce and rehash (i e. no originality); 2) is unable to check (or even interested in checking) its output against external reality.
I guess it’s good for creating value plausible bullshit. But what else?
It’s impressive as a technical feat, but I see no usefulness. Sort of like crypto.
@Kurtz: That was the funniest thing I’ve read today, or maybe this week. This year so far is in the running, honestly.
@Michael Reynolds:
At some point someone will add a Zwicky box of potential human experiences to the mix.
@Just nutha:
Yeah, I know. I should have said it was more literate than Donald Trump’s tweets.
No, what’s going to be really interesting is to get a ChatBot Detector AI and run it against the ChatBot AI.
Then develop functionality can mod ChatBot AI each time Detector flags it, and Detector AI each time Detector misses it, integrate with both.
And iterate at maximum speed.
Wheee!
🙂
Probably even better to occasionally get a human sanity check, LOL.
It needs a thesaurus. For example it uses speculation/speculate over and over.
@Stormy Dragon:
Thanks,
I’m still baffled as to why such a ship carries so many conjoined twins.
As I recall in The Pickwick Papers, a character gives an address about Chinese economics by reading the encyclopedia article on China and the one on economics and combining the two without actually knowing anything about either. This is a biological form of artificial intelligence.
Considering the information the ChatGPT gave you is a couple of days old, that tells us something about how often it updates its database or sources.
A lot of people are poo-poohing it, but I see a lot of potential future applications:
– It could be a great substitute for Google and Wikipedia which provides the ability to provide tailored, custom summaries for basic research purposes. Some examples:
-Hey AI, give me a 500 word summary of developments in the South China sea over the last 30 years.
– Hey AI, give me an English summary of news stories in Russian-language news outlets over the last 24 hours with linked citations at the end.
In short, it has the potential to make certain types of information and research a lot more accessible and useful.
I say we feed the AI the US tax code and then let it settle interpretive disputes!
@Kurtz‘s example has me thinking that it could develop into something like photoshop for creative writers. The writer takes some ideas, filters them through some AI profiles, and tweaks the final result. Creative control is still with the writer, but the AI becomes a tool. The photography profession has mostly gotten past the idea of manipulating photos, chances are the writing profession will eventually do the same.
For more academic writers and journalists it could take over part of the writing portion that is drudgery. Give it some facts the journalist collects about a story and let the AI assemble the first draft. Think of how many STEM doctoral dissertations could be improved! 😀
And if a publication writes with a particular style of voice, then AI could change the writer’s natural style to match the publication’s.
In short, I see this technology as mostly a complementary one – at least once it gets better and more diverse in terms of capabilities.
@Stormy Dragon: I was on a completely different wave length. I was thinking a spaceship composed of an LPG tank equipped with a propulsion device. But my vision would explain why all the beings were trapped in the LPG tank more readily. (Or maybe I was misreading. I was working in a SPED class today where the middle school student cohort was playing Kahoot! using “Grade One sight words” as the game content.)
@Andy: “– Hey AI, give me an English summary of news stories in Russian-language news outlets over the last 24 hours with linked citations at the end.”
My understanding is that the AI can’t do the documentation task to format requirements yet, but that’s not a high bar to clear. And yeah, that would even be valuable to someone like me who doesn’t need an AI to write my terminal product.
@Kathy:
I assume MR meant it as a reference to:
https://en.wikipedia.org/wiki/Sea_Org
In universe, maybe Zoe was able to pick up an old LGP ship cheap from a cracking yard and use it to house her growing cult.
I should have made the connection earlier.
One of Asimov’s robot stories deals with a proofreading robot. The title is Galley Slave, and it dates from 1957. The gist is a product liability lawsuit, where a scholar contends the robot in question, EZ-27, rewrote a part of his book, which made him look incompetent and thus damaged his reputation.
SPOILER ALERT
At the end, the scholar in question claims using AI extensively will reduce the role of scholars and researchers to instructing robots and computers. This goes well beyond proofreading, of course, but he saw the EZ-27 unit as the thin edge of the wedge.
I have long wished I could just sit back and come up with concepts, and be paid for it. Unfortunately, that’s not how fiction works – the big idea is easy, it’s the ideas needed on every page that wear you out. Pity. At this level I’d have to babysit ChatGPT like a mediocre ghostwriter.
@Michael Reynolds:
Right. I recall a story you related about doing research for a heist novel. Just from that tiny snapshot, I pictured your avatar (yes, I know, your pic helps) skulking around a European museum, drawing the attention of security guards.
After I entered the prompt you generated, I simply asked ChatGPT to “tell me a story” and it was more like a plot outline than an actual story.
A child loves his little town and all the people in it. He is for some unstated reason known for his bravery. A dragon begins menacing his town. The brave child slays the dragon. How? We don’t know. We just know he loved his town so much that he killed a dragon with nothing more than the word brave.
ChatGPT can sort of tell, but it can’t show. Is it possible that a user can coax a show out by iterating? Maybe. I can’t rule that out, it may develop along those lines, sort of like how @Andy describes the way photographers got better at using Photoshop.
But I am much more skeptical that it will be able to produce interesting writing, a sui generis voice, even if it can generate more detailed plots. Then again, if we trained it on, say, the text of three of your books, it may be able to do an impression of your voice. But I suspect it would contain some artifacting.*
Further, at least for the time being, I don’t see it generating any kind of story that would trigger a compelling moving picture in my head.
But it’s early yet.
*actually, it may be more interesting to feed it your OTB comments and start asking it political questions. But that may be a little like dropping six Orange Sunshines and finding the largest mirror in town.
I see ChatGTP as a big threat to the non-exceptional white collar creative class. I’m thinking the writers for the average SitCom and Soap Operas, plus B-C level movies (Hallmark). I do not think the LA Entertainment industry understands they are on borrowed time – then AI not only replaces the writers but also the characters in the relatively cheap Hallmark movies where folks th do 40 movies per year. I welcome Eddie and Michael’s comments – I’m not in your business but I know a few (3) in the documentary world (as in Hollywood, not Indy stuff) from High School – today producers.
This past winter I was involved with an infirm neighbor who was addicted to the Hallmark Channel – it would be very easy to automate in AI.
However: Need for celebrity.
Growing up a good friend’s dad was a writer for The Waltons, on a manual typewriter, and I knew Grandpa Walton was a gay commie (this was the late 70s). “Goodnight john Boy”