Comment by kleton | Hacker News Reader

Comment by kleton | Hacker News Reader

This kind of filler-laden writing will be quickly eradicated, like the cockroach, but by AI. I had bing or copilot, whatever microsoft is calling it in the edge sidebar answer the question I clicked the bait for.

> According to the article, they solved the cockroach infestation by using a new product called Combat, which was a bait that contained a slow-acting poison that could kill roaches through various mechanisms of transfer, such as feces, vomit, cannibalism, and necrophagy. The bait was attractive to roaches and could wipe out entire colonies, even if only a fraction of them ate it. The product was developed by American Cyanamid, a chemical company, and tested by Austin Frishman, a cockroach expert. The product was launched in 1985 and became a huge success, reducing cockroach complaints and populations across the country. The article claims that this was a forgotten achievement in the history of pest control, and that the roaches are slowly coming back due to behavioral resistance to the bait.

Source: Conversation with Bing, 12/27/2023 (1) The Cockroach Cure - The Atlantic. https://www.theatlantic.com/podcasts/archive/2023/11/cockroa.... (2) Cockroach Infestation - Signs, Prevention, and Control of Roaches. https://www.familyhandyman.com/article/cockroach-infestation.... (3) How To Get Rid Of Roaches - This Old House. https://www.thisoldhouse.com/pest-control/reviews/how-to-get.... (4) How To Get Rid Of Roaches – Forbes Home. https://www.forbes.com/home-improvement/pest-control/how-to-....

>This kind of filler-laden writing will be quickly eradicated

It's a transcript of a podcast. I agree it was annoying to read, but the purpose of a podcast is not maximum-efficiency transmission of information.

All good conversations ebb and flow, and the interview process is just trying to distill the conversation to some entertaining or useful end. Some would use A.I. in an attempt to distill it more, and may find some gem or morsel worth consuming, but miss the depth and value of the conversation itself. It may be time saving for information transfer, but at the expense of the connection that a conversation grants the participants (and to a slightly lesser degree listeners).

Remember that this technology was developed for the NSA because of the data deluge it's surveillance systems created. Who had time to listen to hours of recorded phone calls looking for potential dangerous activity? Speech to text helped alleviate the problem for a short while, but when the internet took away the monopoly the phone companies had on long-distance calls, the volume of international calls went ballistic, and the NSA needed a way of filtering the data to something more useful. In comes A.I. to the rescue!

Who wants to read a transcript of a podcast. 99% invisible does this so much better - a long podcast with all the details, accompanied by a short article with photos. More work of course, but that's what the AI solves.

People who can't hear, just for example?

And fwiw, "this isn't condensed enough" could be applied to 100% of novels, achieving a 100% compression rate, since none of it is useful.

I'm very sorry if you're the one person with a cockroach infestation that has to be eradicated right now and you critically depend on the information in this one article and don't have access to any other resources, but maybe just accept that there are different writing styles for different things, audiences, and even over time for the same person.

> People who can't hear, just for example?

I don't see why people who can't hear would like to read a transcript of a podcast rather than an article either.

I understand what you mean, but if you subscribe to the Atlantic it’s because you actually want long form writing from some of the best essayists of our time.

It’s not for everyone. Especially when our attention spans are shorter than ever.

To be fair, this one is a transcription of a podcast. It's not really the kind of long form writing Atlantic is famous for, so I can understand the GP wanting a shorter version.

While I will certainly find AI summarization useful I think in this case the summary misses the human experiential component that we get from the recollections of the host and guest.

Precisely what I don’t care to read. Especially when they start the article giving background, tease that there’s an important thing coming, and then instead of getting to the point, it starts an interview which resets the conversation back to talking about the background again. I absolutely hated everything about the way this article was written.

Might you share what your question was? Or did you simply say "summarize"

It was entertaining to read.

I agree about the writing, and I gave up on the article.

How do you know it's accurate? How do you know which details are accurate? For example, was it called Combat, or was that another product in the article (maybe a failed one)? What important information was omitted?

Issues like that make the summary useless, IMHO. (Not a criticism, but a point about the value of information with unknown accuracy.)

AI summaries have been common on Reddit for almost a decade now. Where does the problem of "accuracy" come in? They weren't even using LLMs

How does that address the issue? Being common on Reddit (or any social media) isn't evidence of a good idea. It could be just as accurate or inaccurate on Reddit, for years.

> "accuracy"

How could there be scare quotes around accuracy? You question its reality or importance?

Where is there ANY evidence that these summaries aren't accurate? Until you produce it, it's as substantial as ghosts and evil spirits.

The burden of proof is the reverse, in any field of knowledge it's on the person making the claim. I could make up anything in the world (and many people do); it's the requirement for support that separates truth from fiction, science from fraud, fact from misinformation. You can't publish a scientific paper without evidence, and say 'it's true until someone proves otherwise'; it's false until you prove it. Same in a courtroom or anywhere else.

Until there is sufficient evidence that the summary is accurate, there's no reason to believe that it is.

fwiw, all caps is against HN guidelines. I'm mentioning it because there seems to be a rash of it this week.

> The burden of proof is the reverse, in any field of knowledge it's on the person making the claim.

You made an unprovoked claim of inaccuracy based on what?

The universal, clever retort! Let's agree then that nothing is demonstrated without sufficient evidence.

I think it's a transcript of a podcast, to be fair. But yeah I had the same reaction.

Reply on news.ycombinator.com