Just got the “UHK 80 reviews from around the world” newsletter delivered to my mailbox. The content was overall something I’d want to read (what a handful of YouTube reviewers had to say about the UHK80), but the presentation was tragically AI.
This is purely my own opinion, but it was an absolute disappointment to read. I altogether stopped reading partway through, due entirely to how AI-written it was. A real letdown from the UHK blogposts I’ve been accustomed too. Those were authentic and sincere, clearly written by Laci. They’d always been a treat to read. Which made this newsletter all the more disappointing. Full of AI text. Mind-numbingly same-y.
Again, I want to make clear that this is only my own opinion. Not everyone hates the stink of AI text. But for me, it’s not only an aesthetic issue (ugly prose) but also a respect issue: if a human doesn’t have time to write it, I don’t have time to read it. The world (the internet) already has more information than I could hope to read in a thousand lifetimes–yet we human beings have less time than ever before. So it’s a horrible waste to spend our limited minutes and hours on text that was spit out from an LLM.
I don’t want to give you too hard a time, Laci. I’m so supremely happy with my UH60v2 that I use every single day. It’s a superb piece of equipment, and the computer-y world is better for you having invented the UHK. But, speaking again just for myself, this kind of writing is revolting. Alienating, even. If that’s a sample of future newsletters and blogposts, I’ll certainly be unsubscribing. Life is too short for me to read text that’s not written by other humans.
Claude indeed generated the per-review summaries for this newsletter from the YouTube captions. I read the summaries and thought that they capture the essence of the reviews very well. Then I made a couple of corrections and concluded that I couldn’t write better summaries as a non-native English speaker. Finally, I had some of my friends review the newsletter, and I published it.
However, even though I think it was a sensible solution for this content, we surely want to write and publish authentic content going forward.
Please always speak your mind regarding such cases. Our newsletter should provide value, and if LLM-generated content, including occasional content, is unwanted, then I won’t use LLMs for such purposes at all.
Lastly, thank you for the kind words! Glad to hear you’re happy with your UHK!
Thanks for such a gracious response. I think that speaks volumes about you and that kind of attitude trickles down throughout the entire community on this forum. Thanks.
On being a non-native English speaker, I’m totally sympathetic. And even as a native English speaker myself, I can imagine that writing video summaries would be (at best) boring and unrewarding. The prospect of using an LLM to cut that out would be very appealing.
Thanks again for being so receptive to my criticism.
Oof, that was a bit harsh, IMO. Did you at least skim through the text, and watch the videos? I think that was the intended result, at least.
I admit that AI-everything tends to annoy me as well, but I really don’t see a problem in this instance. After all, it did point me toward a few decent videos, crafted by very real humans (that I likely wouldn’t have seen before reading that blog post BTW). People have been asking for video reviews, and Laszlo did his best to curate a few.
Anyway, I think it’s important to keep in mind just how incredibly complex the UHK stuff is. I’m honestly in awe of how well it’s being managed by only a handful of (probably super exhausted) workaholics.
@mlac Maybe it would ease things a bit if you stamped an AI TL;DR disclaimer on such segments?
Insofar as it was harsh, I meant it to be so. Not as a judgment against Laci, the UHK itself, or the UHK project as a whole. But I absolutely meant it to be a judgment against this mailing.
To answer your question, yes, I did skim. I did try to read. Eventually, I’ll watch the videos just like I’ve watched every single review video Laci has ever featured on the blog. Like @Tunney, I have always enthusiastically read everything that comes from UHK. It’d always been good before. That’s what made this such a disappointment.
The possible benefit you point out (of being pointed towards some decent videos) could’ve been accomplished even more easily without any AI, simply with a bulleted list of links. Or, if we’re feeling fancy, with a 1-2 sentence, human-written blurb about each video.
I think you missed some of what I was saying: overall there was content from this post that I did want (i.e. being pointed to videos about the UHK). It’s the AI-generated presentation that I objected to. At no point have I lost sight how hard it must be to run this business and make these great products. But that has nothing to do with what I think about AI-generated prose in my inbox.
I didn’t mean to be too hard on anyone here either. Everyone’s entitled to their opinion, but spoodle’s reaction caught me off guard.
I guess I just see the use of AI in that blog post to be an acceptable and effective use of a tool. I’m a musician, and I can agree that AI used in the arts is often gross and quite creepy, but I didn’t feel that was so in this case. They don’t all have to be prosaic masterpieces. Just my opinion. But if people have such a strong reaction against AI, so be it.
I feel that the entire thing is best addressed with simple transparency. I think just heading the section with “AI generated summary” would have made all the difference, as @pcooke9 mentioned.
I write mostly to say that I respect the constructive reception by @mlac of the criticism and to second the above sentiment that transparency is a good middle ground.
