oliver2002

Blog authors have been using templates to write articles since ages... remember the accidientially published template by MMS? The ChatGPT generated blog posts follow a similar pattern.

What seems to be of concern for websites like Reddit is that AI content generators datamine their sites to gather information. Reddit is going to charge such commerical users to mine the info in their website.

Some insight is in this article: https://www.marketplace.org/2023/04/...nies-for-data/

GUWonder

Using templates is standard in the legal, financial and customer service worlds too. [Why pay more to reinvent the wheel in any sector/industry unless the reinvention/learning is part and parcel of fleecing clients/customers/audience? ] And computer algorithms generating content in line with background templates is also the expected being realized.

I strongly suspect that one or more of the AI bots have already mined FT too. We also know that the miles/points blogosphere have been mining FT even before "AI-generated content" went so "mainstream" as to be part of the general public discourse. How are they going to restrict that "traditional" mining being done?

Given government regulation tends to trail technological development or be sloppy when rushed to try to "catch up", expect the wild west for a while and no guarantees that the likes of Reddit succeed to do more than charge the straight shooters to get "updated" content. Reddit already has battened down the hatches in some ways even well before AI-generated content went "mainstream" as a talking point, but it's not to the point where enterprising third parties and government actors wouldn't be able to get the content anyway and package it for inclusion in large language modules.

CLEguy

Better for whom?

DELee

The algorithms.

David

GUWonder

For those in favor of advancing statist and corporate-friendly narratives under the disguise of neutrality and/or objectivity with boring language choice/structure in public. The technology ends up favoring the “powerful” at the expense of the others and of challenges to the narrative of the “powerful”/“in-crowd”.

Just wait until the miles/points sphere comments get moderated by these AI tools or that these AI tools end up being used to enable “super moderation” by a more limited/concentrated set of comment moderators. Nothing corrupts more than more concentrated power within and over a broader portfolio.

Will be interesting to see if any of the existing sites in the space that have shut down commenting will decide to open them again while putting the “burden” of “content moderation” of the comments on AI tools.

GUWonder

So Reddit — including in the travel related sections — is at times having content removed under some kind of rule about “AI-Generated Content are Restricted”. Not always so amusingly, content can be falsely flagged as AI-generated content — something that students at schools are sometimes even finding out after graduation — and there isn’t necessarily a very timely and universally-accepted remedy to the implications of false accusations about the source of the submitted content.

Have any of the other travel/miles/points sites yet adopted such a (Sub?)Reddit-like policy too and started to use it with regard to user-submitted comments? If the objective is to disappear some user-submitted content and even to try to disappear some site users, this kind of content restriction policy provides an exploitable and potentially very user-abusive means for the site owners/operators and/or their favorites with administrative tools to do just that.

kokonutz

It's become pretty easy to spot ChatGPT content that isn't heavily edited/re-engineered. The patterns and flow are very distinctive. I see it all over the place now: blogs, social media posts, even here on FT from time to time.

GUWonder

Can ChatGPT even accurately spot its own generated content as being its own? A lot of people around have a lot of confidence in their own ability to recognize ChatGPT-generated content and some even enough so that they are trying to school others into how to try to recognize such content to then exclude; but confidence alone doesn't make for reliability in separating out what is and is not ChatGPT-generated content.

The most distinctive thing about AI-generated content which I've confirmed to be ChatGPT-generated content is how often it seems to peddle misinformation by way of piecing together content from the wrong set of "facts" from that which was asked of it.

As ChatGPT and the like gets increasingly used and the product consumed by readers, it's possible that more and more readers will end up having their own language use subject to modification following from longer term exposure to such tools' output. Won't that make it harder for a casual reader of the content to recognize what is and is not computer algorithm-generated content?