Is this the future of media?

[MichaelB]

CNET announced (here) it is using AI to write articles for CNET Money. In the past 2 months 79 articles have been written by an AI bot, probably ChatGPT. CNET says they are all fact checked and reviewed by a human editor, but other media sources report numerous errors. The announcement by CNET came only after the website Futurism first broke the story. Looking over CNET, it is difficult to distinguish between AI generated vs human written content. This is mostly because the articles are bland and mediocre, with no real investigation or analysis. Here are articles by The Verge (here) and Gizmodo (here) with more details.

The same thing is (probably) happening with content on The Points Guy and Bankrate.com. All three were bought by the same private equity company and this is how they monetize. Of course, all the stories have affiliate links.

So, is this the new normal for media? ER Forum members love to link and discuss financial articles, especially when the authors don’t seem well informed or knowledgeable. It that is the case, we will have a gold mine ahead - an endless stream of articles that seems to be little more than a rehashing of content available elsewhere.

It is now more important than ever to examine the sources of information we consume, not because of partisanship or political leaning, but to ensure they are not just fronts with hidden interests or financial agendas. The real risk here isn’t just the existence of hidden interests, it’s that an AI bot can churn out a literal endless stream of content that can overwhelm search engines and completely push aside real analysis. Selecting one’s sources becomes more important but also more challenging.

 

[bloom2708]

ChatGP overload and it is only starting.

What college professor would assign papers when ChatGP can write the paper so easily.

All exams should be spoken presentations with only a note card allowed.

With AI and self driving cars, we can just sit in our power chairs like in Wall-E. One step forward and 2 steps back.

 

[audreyh1]

Yuck!

More reason not to read general internet articles I suppose. A lot of simplistic financial articles already read like a bot wrote them

I follow certain sources via Apple News and a couple of other subscriptions. I guess I’ll have to keep an eye out for this.

 

[W2R]

The real risk here isn’t just the existence of hidden interests, it’s that an AI bot can churn out a literal endless stream of content

I'm not lighting my hair on fire and running for the hills...

Articles could be "endlessly" written by less sophisticated AI back in the mid 1980's, when I took my first class in AI back in engineering school. But is there in fact a vast craving for such articles, or will most people just yawn and go elsewhere for their information? My guess is the latter, if they have a choice.

Yuck!

More reason not to read general internet articles I suppose. A lot of simplistic financial articles already read like a bot wrote them[...]I guess I’ll have to keep an eye out for this.

I suspect that those articles, and much of the content we read online today (and in past years) was/were already produced by AI, in whole or in part.

 

[Qs Laptop]

NY Times had a test a couple weeks back where they had 12 writing samples and asked the reader to guess which ones were written by humans and which ones were written by AI. I only got one incorrect.

I went to chatGPT, created an account and tried it out. While it is somewhat impressive I can tell it was written by an AI. Also, it got two of my questions factually wrong. One of the questions it got wrong was, "Explain the infield fly rule in baseball."

This is the wave of the future, no doubt about it. Right now AI writers are in their infancy. They will only get better. As Audreyh1 says, a lot of financial articles read like they were written like a bot. That is because they have been written by a bot. I'd say for at least 5 years now. Same thing with many news stories.

 

[audreyh1]

As Audreyh1 says, a lot of financial articles read like they were written like a bot. That is because they have been written by a bot. I'd say for at least 5 years now. Same thing with many news stories.

And they still are given headlines that don’t match the article content, even a direct contradiction at times.

 

[ExFlyBoy5]

ChatGP overload and it is only starting.

What college professor would assign papers when ChatGP can write the paper so easily.

I don't know...that could spell trouble since ChatGP *loves* to plagiarize and it is terrible at citing sources. Great way to fail a course and perhaps get kicked out.

 

[W2R]

I don't know...that could spell trouble since ChatGP *loves* to plagiarize and it is terrible at citing sources. Great was to fail a course and perhaps get kicked out.

Science fiction in the 1960's was simply FULL of terrifying stories about how AI was going to take over our communications and society in general. But somehow we got through the 1970's reasonably unscathed.

https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/

 

[athena53]

Yeah, I've seen that. They tend to use a ton of fluffy padding so there's more room for ads. Go on and on explaining why the topic is important, why everyone is wondering about it, the history, etc. Preface sentences with "You see,..." and other useless words. Make suggestions that include product links.

Just this past week I clicked on an article on things retirees regretted and supposedly something like 30% of them regretted not buying an annuity. If that's true (and of course there's no source cited), they must have had a greatly exaggerated notion of what an annuity would have paid out for whatever they could pay in.

 

[FLSUnFIRE]

AI may be able to write articles simply gathering info from other sources and repeating it but any creation of new ideas/articles is a ways off IMO (unless we are all in a sim and every author is AI?). Steve Lehto had a pretty entertaining video on an AI lawyer recently.

 

[easysurfer]

Would surprise me as there seems this cultural obsession with automation.

From the Roomba to self-driving cars and many in between. So why not AI articles?

 

[marko]

On a positive note, might we eventually end up with articles that are more factual and less agenda/bias driven?

 

[donheff]

Yeah, I just asked it what the best SWR for retirees to use and got an answer that could have appeared in any money advice column:

"The safe withdrawal rate (SWR) is the amount of money that can be withdrawn from a retirement savings account each year without depleting the account over the retiree's lifetime. The most commonly used SWR is the 4% rule, which states that retirees can withdraw 4% of their initial savings balance in the first year of retirement, and then adjust for inflation in subsequent years. However, this rule is based on historical data and can vary depending on various factors such as the retiree's age, portfolio allocation, and expected future returns. It is best for retirees to consult with a financial advisor to determine an appropriate SWR for their individual situation."

Edit: I didn't initially realize you could string it out into a conversation. I then asked it "So, what if my bond allocation is 80%. It gave me a pretty detailed and good three paragraph explanation and caution.

Edit, edit: This thing is loads of fun. It gave me lots of accurate info on DIY indexing vs active management but it will not be pinned down on what is best. Always ends with a recommendation to consult with an FA.

 

[pb4uski]

Sounds like a good tool to do basic research on a topic.

 

[donheff]

I am meeting with my GP for an annual physical on Monday so I asked it a bunch of questions I planned to discuss and got a lot of things to pursue further with my doc. This thing is pretty darn good. Probably a useful first step on Internet research. It can give you an instant overview of your issue that can then lead to more questions and ultimately independent fact checking.

 

[JoeWras]

On a positive note, might we eventually end up with articles that are more factual and less agenda/bias driven?

Uh, I hope so, but right now, no.

I've been following the "debt showdown" issue and it seems that some of the AI bots have regurgitated biases gathered from other writers as fact.

What I'm saying is: clear speculation seems to sometimes get turned into fact by the less sophisticated AI writers (which mostly just copy and pass along).

This may be a feature and not a bug in that some of the outlets want that bias to be loudly proclaimed.

 

[Qs Laptop]

Sounds like a good tool to do basic research on a topic.

Yes, which is why Google is terrified of this thing.

 

[folivier]

And they still are given headlines that don’t match the article content, even a direct contradiction at times.

So true. Especially on news articles. But mainstream tv media talking heads have been doing that for years. Unless they're robots.

 

[Midpack]

It will be interesting to watch this play out. I think AI will become an amazing force for good and unfortunately bad, but it will take longer than some prognosticators seem to think - just like full self driving cars that seemed right around the corner (I was fooled). It’ll happen, but not soon.

 

[bizlady]

New "drawing" Ai out too getting buzz. You can create art or images. Same company, same login and password
https://openai.com/dall-e-2/

Combine the two (text and image) and I suppose you can write a book!

 

[donheff]

New "drawing" Ai out too getting buzz. You can create art or images. Same company, same login and password
https://openai.com/dall-e-2/

Combine the two (text and image) and I suppose you can write a book!

The Washington Post has an article about a guy who did just that and is selling his resulting children’s book. He is getting both cheered and vilified, as is the guy who developed the art ai app.

 

[Mr._Graybeard]

AI may replace humanity, but first it needs a little fine tuning.
CNET is now letting an AI write articles for its site. The problem? It's kind of a moron.

 

[Qs Laptop]

AI may replace humanity, but first it needs a little fine tuning.
CNET is now letting an AI write articles for its site. The problem? It's kind of a moron.

Yes, that is a good example. I also saw one on Tom's Hardware (high tech DIY website) that asked the chat bot how to build your own computer. The chatbot correctly instructed the computer builder to put heat sink paste between the CPU and the heatsink fan, but never mind about putting heat sink paste between the CPU and the CPU socket plate. Also didn't mention removing the retaining arms on the CPU socket prior to installing the CPU.

https://www.tomshardware.com/news/chatgpt-told-me-break-my-cpu

 

[sengsational]

On a positive note, might we eventually end up with articles that are more factual and less agenda/bias driven?

Now the real writers will be writing to manipulate the AI and let the AI take the heat for being biased

 

[audreyh1]

Yeah, I've seen that. They tend to use a ton of fluffy padding so there's more room for ads. Go on and on explaining why the topic is important, why everyone is wondering about it, the history, etc. Preface sentences with "You see,..." and other useless words. Make suggestions that include product links.

Makes me think of the recipe sites where many recipes are extended blogs (going on and on and on) that have lots of rooms for ads and product referrals.