AI Slop and The Question of Authenticity

Right now it's easy to stumble across websites, videos, music, and social media posts that were written by an LLM. Coming across these posts, it's easy to feel as if somebody is trying to deceive you - they want you to click on their content so they can receive ad revenue, and once you see that it's AI-generated, you click away, but by then they've already earned their money.

Part of the issue is that internet platforms are an attention economy, so clicks and views generate revenue. For those living in countries with relatively weak currencies, maybe it's possible to live off $5 of ad revenue per day [TODO: verify]. Especially since 2020, the internet has become a noticeably more global place, and maybe the change in content comes as a consequence of this. This is more a diagnosis of the cause than a call to action: innovation, rather than regulation, will probably lead to a better solution. What we need is a technology that filters AI slop, not some kind of digital ID system that verifies your humanity.

...

There's a popular counterargument that goes something like this: many other technologies were initially unpopular, but grew normalized over time, so the early critics appear to be fools in hindsight. Take electronic music, for example. People may have decried it because it lacks a "human touch," but now there's a massive body of electronic music [etc etc] by artists like Aphex Twin and Depeche Mode. AI slop is no different: one day, people will read novels written by LLMs - instead of Nabokov, people will read Claude; students and academics will produce papers about Claude's writing style and "Weltanschauung," using another LLM for assistance.

However, this argument [possibly a strawman] misses a crucial point: [TODO: generalize from electronic music] the best electronic music doesn't try to imitate analog music, but [embraces] uniquely electronic qualities. It does something that couldn't be done before. In contrast, AI slop is merely an imitation of media that already exists; it just takes less time to produce than it would using a more traditional method.

is media a way for humans to communicate with each other? does it

make a difference to people if there's nobody on the other end?

speculation, not argument.

allegory of the cave: difference between essence and appearance -

but we are not compelled to be platonists

where AI could be genuinely useful, à la electronic music

...

Proponents of AI come off as obnoxious, even aggressive. The classic refrain of "AI will take your job." Statements like "," while less threatening, still come off as condescending impositions. "It's inevitable," they say. You may not like this technology, but it will rule every aspect of your life in 5 years. Disregard the fact that OpenAI[1] is in billions of $ worth of debt, and the massive AI push by FAANG companies may come from desperation to make good on their gargantuan investment. Many AI advocates are probably entrepreneurs, and are enclined to embrace the "next big thing." Nobody participating in this discussion is without bias.

Detractors not much better, bluesky/mastodong type stuff, silly arguments like "AI is fascist," "AI produces too much CO2," and calls for regulation

In 10 years, the hype will have died down, and generative AI will remain where it's useful. People will have adapted to it. The same is already true of technologies like blockchain and [TODO]. Maybe the best approach right now is to experiment with this technology, try to figure out the best way to use it, but not fall for the hype. And also, don't get tricked into supporting regulation

The entire AI debate has grown old, and it's a shame that it's taken up so much space in discourse - both online and off. In fact, I feel guilty writing a post about it. But if you read this far, it can't have been that bad, right?

--

[1] What do they mean by "open," anyway?