Hi Srajan, I hope you are very well. Earlier this year I had an unexpectedly interesting experience using ChatGPT to help assemble a used system for my wife. What stood out was not just product matching but how well it seemed to understand my listening preferences and system synergy. That led me to see if I could build a site that offers a similar experience grounded explicitly in audio and system synergy. It uses AI lightly for language and synthesis but the underlying logic and data are built from structured work I've done using my own experience alongside published reviews and manufacturer philosophy including reviews from you, John Darko and Twittering Machines. I've attached screenshot examples to give you a sense of what it can do. I would greatly value any feedback you have. Thank you, look forward to hearing your thoughts, Mike
Hello Mike, using AI is simply the next step of aggregation which uses the creative content of others to synthesize it in condensed albeit interpreted form. As a creator of original content, I nearly consider such an attempt repackaged theft by someone unwilling or incapable to create their own original content. And to be sure, it seems perfectly legal. My response isn't about that. It's about the difference between creating something from nothing by doing the actual work of reviewing; or converting the work others have done into a summation or excerpt which is additionally prone to assumptive errors. Srajan
Hi Srajan, understood and that's the opposite of what I want. Everything is/will be sourced with links back to your site/sources it draws from. But you're raising an interesting concern especially if it should limit the amount of traffic in some way to your and other reviewers' sites. Are citations and referrals enough, or is there an additional way to recognize the value of the original work behind a site like this? I would want this to be fair and I would like to explore the best way to arrange this. I also want to emphasize that what I'm trying to build is closer to a listening and system advisor: something that helps people identify their own preferences, understand system interaction, and avoid unnecessary churn. Thank you for your response, warm regards, Mike
My comment wasn't really about formal credit. It's about first doing the wrong thing then apologizing for it to make it okay; or in fact, first apologizing then doing it anyway. That still doesn't make it right. YouTube will shut down or demonetize any video in which unlicensed background music exceeds 10 seconds for breach of copyright. My content is likewise under copyright. And I'm giving nobody permission to use it for their own ends. Practically of course, it's been done all along. I'm only saying that from my perspective as a creator of original not rehashed content, this type thing is predatory as in, "the content is out there, it's unprotected, let's repurpose it". Secondly, there's big error potential whereby the end result will be presented as "based on what Srajan said" whereas, when asked about said topic, I might advise something else altogether. Again, it's not very creative but predatory and, strictly speaking, in violation of copyright. So I'm not authorizing it. Legally though it's probably naught I could enforce because if you simply rephrase my findings in AI's words, it's no blatant copy 'n' paste but a more veiled processing and alteration to stand on its own whilst riding on my credibility, findings and reputation. I've worked on those aspects for 24 solid years. You'd start with zero exactly as I did on day 1. But with your scheme, you'd actually ride my reputation—or that of anyone else whose work you harvest. From where I sit, that's lazy and wrong. But if that's what you feel compelled to do, the age of AI is all about it. Take a lousy photo, use AI to make it look splendid; or convert it to look like a Rembrandt painting then think to yourself, "whoa, I painted that". Cheers. Srajan
Hi Srajan, thank you for the thoughtful response. Just so you know, I'm not interested in building something that people whose work I genuinely respect would view as exploitative or harmful. Your perspective is helpful to me even if it's difficult to hear. What I'm increasingly realizing is that the better direction may actually be around listener preference, system interaction, trade-offs and reducing unnecessary churn—not compressing or replacing reviews. Thank you again for the honesty and candor. Warm regards, Mike
Hello Mike: You asked for my opinion. It's what I gave you. What you do with it is entirely up to you. If you decide to go forward, I simply want it on record that I didn't agree to have my content harvested in this fashion. It'll have to be an unauthorized venture. Only people who do what I do know the serious and constant amount of work which goes into it. I dare say that you do not know. How could you? You haven't done it. Having AI reformat and rephrase my work in a matter of seconds is something anyone can do by simply directing Gemini at my site armed with a particular query. It becomes an accelerated highly customizable search function. I have nothing against that on a one-on-one basis. It applies a smart algorithm to search a large data base according to user-definable parameters; including predictions on how certain gear may interact even though the monster variable of the room remains unquantified. Fine. But what you're proposing isn't one individual doing this for themselves. You're proposing to turn it into its own entity. Sorry, not with my blessings. The reason is very simple. I have no control over a final result that associates itself with my work. For the same reason 6moons.china never happened. I couldn't read the translations to ascertain that they reflected original meaning and intent, never mind witness behind-the-scenes interactions to procure/return loaners and the review process itself in a different country. With my existing syndication agreements which compensate the other party for their copyright use and collaborate with individuals whom I have personally met, everything is in its proper place so works as expected. Srajan
From John Darko: I spent most of yesterday dealing with a serious website tech issue. Thankfully Google Gemini walked me through how to solve it. And that's actually relevant to this topic because it's not the first time I've leaned on Google's AI for tech support. Back in January I used it to set up a new Linux server. Last week it helped me locate a Mac permissions issue that prevented a Motu sound interface from making noise. I also use it to find DAK Coffee dealers in Berlin; and what the hell happened to the Keith Haring artwork that used to stand at Potsdamer Platz. There's the rub: Google scrapes my content but I also take from what it has scraped from others. Complaining about Gemini or any other large-scale AI service scraping my website would be hypocritical. The scale is the key differentiator though. Google, OpenAI and Anthropic scrape at scale but also give back at scale. A smaller hifi-focused AI platform would have me asking how I'd be compensated for my work being scraped. It wouldn't be scale; the hifi community is too small for that. And almost certainly no financial compensation either. From where I am sitting as a publisher of hifi content, it would be all take and very little give. So, naaaah as they say in Australia. Count me out. Mike's AI has to scrape your work to reformat it. And I don't care what the outcome is – short version, long version or written in the style of Shakespeare: I won't allow my work to be scraped without the LLM scale that can give me something back in return. John
