AI generated playlists

Music Room
41

True, it was specific for my needs.

That’s also what I hoped to make clear in the discussion by my remark on the results being better when prompting right.

And I (blame it on my bias) hoped that it would be clear from the descriptions of the playlists that I had been looking for a specific result aimed at the enjoyment of those chains. That it did not, did make clear in the thread though that the majority of the people here are very wary on the use of AI. Probably rightfully so. But I think that we might benefit more from AI than we think is possible.

42

At this point you have absolutely answered your own question i.e. it is clearly not efficient. It is more time consuming and generally less instantly accurate or reliable than every other method available. Neither you or anyone else have provided an example of where there is more efficiency.

There is already ample evidence from the long standing algorithms used by streaming and selling services that such things can’t cope with simple recommendations. Why anyone might think AI might be better at more complex ones is beyond me.

You seem excited by the fact AI gave you some music you didn’t know when that’s an easy thing to source in far quicker ways. Your excitement that it is allegedly music which avoids the weaknesses of your system suggests your time might have been better spent feeding it details of your system and identifying practical improvements.

As regards the things you’ve “discovered” about AI on your quest one can only observe that these are known to most of us here and not news.

I think the headline here might be “devil finds work for idle hands.

43

Well, that’s a conclusion on your end that I don’t see.

Because in ChatGPT I had numerous chats to gain insight on different topics which were never aimed at producing a playlist to emphasize the strenghts of my setup.

It was only at a certain point that I just wondered what it would come up with if I did ask for that. So I prompted for that. And it did come up with playlists that I could not have produced as fast and certainly not with the attention points per track. I realise (just now, forgive me) that you folks have not seen the results regarding the attention points.

I will copy, translate and paste some of them in a next post. I found it fun to listen to the music using these.

So I think there is a gain possible if using the machine correctly. As said, not the holy grail but certainly not as dismissive as you put it.

44

These are some of the attention points as promised. As said, I enjoyed listening like this.

Acoustic Pearls – Track-by-Track Listening Guide

1. Rosa Cedrón – Entre Dous Mares

Notice the cello strings, which are almost tangible in front of you.

The HD650 beautifully showcases the warm resonance of her voice.

Spaciousness: the reverb tail is round and wide – check how long it lasts.

2. Nynke Laverman – Alter

Focus on the percussive breaths and subtle guttural tones.

The Lake People bass here is precise and dry, without being woolly.

The dynamic mini-accents in her phrases come through extremely smoothly yet urgently on the Naim.

3. Agnes Obel – Riverside

Notice the stereo difference between her vocal layers (left/right harmonies).

Piano sound: It’s beautiful to hear how the hammer, string, and body come through separately.

The HD650 gives this track a dreamy softness without loss of detail.

4. Eivør – Trøllabundin

Focus on the more powerful chest resonance versus her whispering upper register.

The HD650 makes the air movement around her voice almost holographic.

Note the deep, dark background space; very quiet yet atmospheric.

5. Ane Brun – These Days

The acoustic guitar has fine string-touch micro-details: fingertips, sliding sound.

Her phrasing is slightly frayed; on your chain, you hear it perfectly and humanly.

Note how far behind her the second voice is.

6. Lhasa de Sela – Con Toda Palabra

Her voice is perfectly centered, yet subtly moving; notice that micro-pendulum.

The percussion is dry and close, a nice test of the Naim’s timing.

The HD650 provides a warm coloration—a real benefit to the atmosphere here.

45

Hmmm…This is a similar debate to so many others at the moment.

I can’t help but feel that much of the issue is down to the issue that what’s being branded AI isn’t really AI at all of course, just a search tool that’s offered and blinged up by the provider with the intent towards more data gathering so they or others can use it to try and sell you something at some point.

1 Like
46

So IIUC from this, the time taken for your specific question followed multiple other questions/data input to ChatGPT, so there was a lot more of your time involved than just asking the playlist question. I wonder if ChatGPT’s memory of your “conversations” will confuse results if some time in the future you change system, then ask the same question. I am sure that it won’t work to instruct it to forget your current system (or anything else you’ve “told” it), so a future question would have to specifically state gear not to include.

1 Like
47

Surely no different to what Spotify already does and has done for years before the AI hype

1 Like
48

Oh dear. That’s one of the saddest things I’ve read in a while. I’m not going to name and shame but you’ve been the victim of a classic example of terrible AI answers.

If you reverse engineer the question, amusingly enough using AI, you discover in about five seconds that it has not been compiled by AI at all. It is a straight rip off of a playlist designed to “test your system”. It was produced by a well known author in a well known hifi magazine and has been reproduced several times over the years on certain audiophile web sites. Certain Google searches will pull it up in their top 10 answers simply because it’s been linked to a fair few times.

I looked back and that’s a playlist I compiled in early 2021 having finally moved to streaming. I’ve labelled it as having coming from a specific magazine but a quick internet search revealed that was not its origin. It didn’t really tell me anything I didn’t know and I didn’t discover anything new off the back of it but I did forget to delete it so that’s done now.

The accompanying text for each is as good an example as you will ever see of AI hallucinating an answer along the lines of what you want to hear.

Again, reverse engineer it by asking for sources. You’ll soon hit the barrier where it says “web sites like” or “magazines like”. Big clue there that it’s citing stuff to sound authoritative but those things are not the source at all. Try to pin it down to sources and eventually it’ll cave and confess.

That doesn’t mean you won’t enjoy the music (Lhasa is always interesting) and nor does it mean that the made up text is necessarily wrong. It’s best guesses are often entertainingly close to accurate because it does simple stuff like understand the key words in reviews of the kit you own and can match that to existing playlists.

Just do a search of “test your system” and you’ll get far better outcomes than AI will ever do.

Worth repeating an earlier point. Never ask AI to help with a question unless it’s a subject you already know well else you’ll have great difficulty parsing the credibility of the answer. Here, you have been framed particularly well.

49

I’m afraid I despise AI in all its forms and think it’s mostly a marketing con.

At best, AI is a cynical way for increasingly desperate IT companies to generate income.

At worst, AI takes away any critical thinking and/or proper analysis of the sources from which the information has been derived which can often be dubious and will lead to increased dehumanisation.

I read an interesting article the other day which asked, with the creation now of so many AI videos, whilst once video evidence used to prove something had happened, will we now have to turn back the clock and only rely on actual eye witnesses?

I know I’m old but I’d willingly travel back in time at least 50, maybe 60 years. As a ‘young person’ said to me the other day, “There’s just too much of everything now.” So maybe there will be some kind of reaction to the increasing digitalisation of humanity. I knew what they meant whilst at the same time the things that people really need are being rationed.

Maybe I’m just a sad old misanthrope but there does seem to some sort of a reaction against having one’s neocortex permanently connected to the Cloud which gives me some hope that I won’t end up as the ‘outsider’ character of some dystopian science fiction film.

Anyhow, each to their own.

Have a nice day!

P.S. And yes, I need to spend less time online!

1 Like
50

This is always an interesting aspect for me. The reality is that the world has always been full of more information than can be contained or understood by a human. The difference now is proximity. Pre internet we’d need to go find it. Now it is allegedly at our fingertips.

We get fed lots of stuff about information overload but that isn’t really a new thing. The new thing is that our filters are broken. Our ability to filter out what we do or don’t need, what is true or false, has malfunctioned in the face of all this proximity.

The premise behind this whole thread is a magnificent example of that. Poster finds new way to eventually find an old thing because he’s seen the new thing, wants to understand it rather than be overwhelmed by it, and tries to find a use case for new thing.

Believes that new thing has found an answer that only new thing could bring but has lost the ability to step back to see that the new thing brought to him by new thing is actually very old thing dressed up as belonging to new thing.

If the OP had put even half as much time into trying to establish where his “AI playlist” came from as they did trying to prove a point to themselves and others this thread would have been in a very different place very quickly.

51

You are right. I know AI comes up with answers that are available on the www. So it will always bring something on the table that is somewhere to be found already.

But that doesn’t change the fact that the output came up faster than I was able to determine by Google searches and scanning all the articles.

So was AI helpful for me? In my opinion it was. Are there other playlists that could be helpful as well? I am pretty sure there are. Could they do the job even better? Perhaps. That does not change the fact that AI came up fast with output that I found helpful and enjoyable.

Yes, but that has nothing to do with the fact that when asked (prompted) it delivers fast. The earlier input may have (probably was) been taken into consideration by the machine, but I never asked to generate a playlist before. The earlier chats were aimed at other purposes so it’s not that that time was just waisted on the playlist. That it does take earlier input into consideration when I did might in fact have been beneficial for my person. It may have come up with another set of songs otherwise (not necessarily worse, but not necessarily better either). The fact remains that it provided a playlist very fast and which I enjoyed listening to.

I guess that sums it up perfectly well. Use of AI is not mandatory. So if one chooses not to use it, that’s fine. And if one chooses to use it that’s fine too I hope.

Anyway I had fun discussing it here and it made clear that I am not a majority (to put it mildly :wink:) in this debate with the view that AI is potentially beneficial where it comes to this subject.

52

No, it didn’t. That is a massive act of self-deception. Once I understood what it was you’d done it took me literally 5 seconds using Duck Duck Go search. No browsing of articles etc.

You, in contrast, spent quite a while with AI feeding it information before it came to the conclusion you describe. There appears to be zero desire on your part to test that against how long an ordinary search engine would take if you did similar in terms of refining your interaction etc. Without doing that you have no idea whether what AI did was quicker or even quick. Measured against what that you have specifically tested? Apparently, nothing.

“It’s fast because it felt fast to me” is not an outcome. It’s not evidenced against anything. It’s an unevidenced, untested assertion based on a false premise. In sociological terms it’s the classic “self-fulfilling prophecy”.

“I found something good so this thing must be good.”

There are many reasons why you’re in the overwhelming minority here. They include asserting something to be an absolute which is true when you’ve neither tested that against anything else nor have any evidence it’s true. They might also include a wilful determination to find a use case where there just isn’t one. “I used AI to do… “ does not equate to “therefore there is a use case.” It requires far far far more than that.

It’s the equivalent of I got from a to b by steam engine therefore there is a case for the steam engine over the electric car.

53

As someone else put it,

“There are some technological developments where as an individual you have to draw a line for yourself. Modern civilization has made it possible to work from home and eat ten thousand calories a day without ever exercising or leaving your apartment, but most of us have the good sense not to do this because we know it would be very bad for our health. We’re going to have to start looking at AI the same way we look at McDonald’s: sure it’s there, but that doesn’t mean you have to consume it, because it’s really not good for you.“

54

Although you stated you did not want to name or shame quite some of your remarks were belittling in tone of voice in my experience. And this last one is not any different. So I just leave it at this.

2 Likes
55

That may be right. At this point I still try to have an open mind on it.

56

Well not yet! Although you wouldn’t know it if you worked where I worked.

57

Apologies, there was no intent on my part to offend. I am just surprised anyone persists with such an obviously unsustainable argument in the face of overwhelming evidence to the contrary. The assertion re: speed was the latest of those.

I’ve a genuine interest in these things, have IT qualifications and have tried this stuff for myself and whilst I”m opposed to so called AI for multiple reasons, I would have been genuinely interested and have paused to self-reflect if the thread had been populated by lots of people going “yeah, i’ve tried that and it’s quick and brilliant”. It hasn’t though and I think the reasons for that are crystal clear. They may be uncomfortable reasons but they remain true.

58

Apologies from my side then as well as I may have misread the messages. English is not my native language and I may fail to pick up on nuances. Let me just say that I did Google again (I also could have used Duck Duck Go) after you said these playlists were so easy to be found but I failed to get the results as quick as you stated. Maybe (or probably) that’s due to my lack of ability with searching in search engines like Google. ChatGPT interacts up to a certain extent. Google doesn’t. And yes, I do realise that that makes it guiding me to a certain direction. As long as I realise that and I feel it still gets me the result than it’s suitable for me.

So therefore I come to the conclusion that at this point in time (given my personal ability to work with the digital tools available) AI worked in this case for me.

I see it as another tool in my toolbox. It’s one of many tools. Like my real toolbox at home. Sometimes I use this tool and another time that tool, the choice depending on if it suits the purpose and my ability to work with a certain tool. But in the end I aim to get the result I want.

2 Likes
59

Indeed it is s tool, and when used appropriately, with awareness of its limitations and how to manage it, it can indeed be useful. However as I have said before, whilst AI can be a great tool when used in an informed manner, crucially as a tool it is only as good as the skill with which someone uses it, and used in an unskilled, uninformed manner, simply believing what it says, is in its way as dangerous as wielding a chainsaw without the awareness and skill needed to use safely and effectively. Of course in a situation like creating a playlist there’s no risk of actual harm or damage so maybe it is a good place for people to play..,!

2 Likes
60

Yes. But the warning about using AI should certainly be heard. I think it starts with awareness of your own vulnerability. Even things that we share because we think it’s harmless could easily be connected to the bigger picture of ones online presence. And therefore be a potential addition to for example a fake profile. So to be careful with it is absolutely true.