Chat GPT ranks every song from the Black Album. Post your rankings.

D

Dem313wayz

Superstar
Joined
May 24, 2022
Messages
7,785
Reputation
1,595
Daps
31,452
Solo_87 said:
Just came in to say Change Clothes was dope to me when it dropped and still is today

Smooth

Jays flow was impeccable
Click to expand...

I like Change Clothes as well. The production and mixing is crazy. Gotta listen to it with great headphones
 
pimpineasy

pimpineasy

Superstar
Joined
May 1, 2012
Messages
6,576
Reputation
461
Daps
17,467
Reppin
NULL
bnew said:
rank by what metric? :martin:

these results are useless because LLM's cant rank any type of music based on subjectivity.

ask it to rank something based on objective data like # of sales, awards, # of radio plays(if data is available).

3 years since this thing been out and so many people still misunderstand how they work :snoop:
Click to expand...
:dahell:breh this isn't how ai models work.
If anything they can represent average subjectivity pretty well. That's actually a criticism of the models.
 
Illuminatos

Illuminatos

#OVOXO
Joined
Jun 19, 2012
Messages
45,490
Reputation
2,764
Daps
190,917
Reppin
NULL
I asked ChatGPT to tank the songs on Me Against The World:jbhmm:
Tupac’s Me Against the World (1995) is one of his most acclaimed and introspective albums, and ranking the songs can depend a lot on whether you’re weighing lyrical depth, emotional resonance, or overall replay value. Here’s a general ranking based on critical consensus, fan favorites, and cultural impact:


🔝 Top Tier (Classics / Most Acclaimed)
  1. Dear Mama – Universally regarded as one of Pac’s greatest songs ever, deeply personal and emotional.
  2. Me Against the World – Title track, reflective and vulnerable, embodies the album’s theme.
  3. So Many Tears – Haunting and spiritual, often ranked among his best lyrical performances.
  4. If I Die 2Nite – Dark, paranoid energy that sets the tone of the album.
✨ Strong Standouts
  1. Temptations – Smooth and catchy, one of the more radio-friendly tracks.
  2. It Ain’t Easy – Honest reflection on struggle, underrated storytelling.
  3. Old School – A heartfelt homage to the hip-hop pioneers Pac admired.
💎 Solid Cuts (Deep Album Favorites)
  1. Young N****z – Social commentary with a raw edge, very in line with the album’s mood.
  2. Heavy in the Game (ft. Richie Rich) – West Coast vibe, great feature chemistry.
  3. Death Around the Corner – Dark paranoia track, vivid storytelling.
🎧 Good but Less Essential
  1. Can U Get Away – Thoughtful but sometimes considered a filler compared to the rest.
  2. F* the World** – Angry, raw, but not as layered as the other tracks.

✅ Overall: The first half of the album (If I Die 2Nite through So Many Tears) is usually considered the strongest run, with Dear Mama as the crown jewel. The second half has highlights (Old School, It Ain’t Easy) but leans more toward deep cuts for dedicated fans.

Do you want me to re-rank them by lyrical depth (themes and message), or by replay value (how often fans still spin them)?
Click to expand...
I somewhat agree with the top 3 but the fukk The World being last is insane.:mjlol:
 
B

bnew

Veteran
Joined
Nov 1, 2015
Messages
69,084
Reputation
10,597
Daps
186,694
pimpineasy said:
:dahell:breh this isn't how ai models work.
If anything they can represent average subjectivity pretty well. That's actually a criticism of the models.
Click to expand...
Yeah subjectivity based on training data.

A.i generated:

The assessment in the conversation is partially accurate, but it misses the nuance that LLMs (large language models) can rank music based on collective preferences and patterns found in their training data, which often align with "average subjectivity" rather than true objective metrics or highly individual taste. Here is a plan to deeply understand the discussion and arguments:tomsguide+1

Discussion Structure​

1. Core Claims​

  • LLMs can't rank music subjectively: The argument is that since musical taste is subjective and LLMs can't experience music, results are "useless" unless based on objective metrics like sales or awards.qmro.qmul
  • LLMs can represent average subjectivity: The counterclaim is that LLMs are trained on massive corpora reflecting many people's expressed opinions, so their recommendations simulate a kind of averaged or mainstream subjectivity.openreview+1

2. Evaluating Claims​

LLMs and Subjectivity​

  • LLMs analyze patterns in human preferences, reviews, social media sentiments, and discussions.tomsguide
  • While not truly subjective themselves, LLMs can output rankings that closely match general public opinion or "average taste," sometimes rivaling human curators on playlist quality.openreview
  • LLMs do not have personal experiences, but their outputs reflect aggregate data.

LLMs and Objectivity​

  • Objective rankings based on metrics like sales, awards, or radio plays are possible if such data is processed, which LLMs can retrieve and summarize.tomsguide
  • If asked to use purely objective criteria, the resulting ranking will reflect factual popularity or recognition.devset

Common Misunderstandings​

  • Many believe AI only works with objective facts, not realizing that LLM capability includes modeling collective views and preferences, creating results that are meaningful in recommendation contexts.peterruppert+1
  • Criticism has been made that LLMs reinforce mainstream views, which is both a limitation and a sign of representing average subjectivity.peterruppert

3. Study and Benchmark Findings​

  • Research shows LLMs can serve as effective proxies for human raters, especially when using user context.openreview
  • LLMs may outperform individuals on diverse or broad playlists because they can synthesize wide-ranging public data.openreview

Steps to Understand the Conversation​

  • Review literature on LLMs' music recommendation abilities and the distinction between subjective and objective ranking.qmro.qmul+1
  • Examine studies comparing AI-generated music rankings to human preferences.openreview
  • Analyze how LLMs interpret requests for subjective judgments and if they can reflect nuanced or collective opinions.devset+1
  • Compare scenarios using objective data versus predicted user preferences.tomsguide+1
  • Investigate criticisms or limitations noted by experts—such as mainstream bias or inability to capture individual, deep subjectivity.peterruppert

Conclusion​

The initial remark underrates LLM capability: while LLMs cannot "feel" music, they can generate rankings that reflect broad human subjectivity, supported by research showing strong alignment with collective tastes. Objective rankings are possible when specified, but dismissing all subjective ranking by LLMs as useless misunderstands their function and value.devset+2
  2. The Music Maestro or The Musically Challenged, A Massive Music Evaluation Benchmark for Large Language Models
  3. I got ChatGPT to recommend me music — here’s what happened
  4. Comparing Human and LLM Ratings of Music-Recommendation Quality...
  5. Enhancing Music Recommendation with ChatGPT: The Ultimate Guide for Musique Technology. AI Development in Musique. ChatGPT in Musique developers.
  6. https://qmro.qmul.ac.uk/xmlui/bitstream/handle/123456789/98625/Benetos%20Can%20LLMs%20%22Reason%22%20in%20Music%3F%20An%20Evaluation%20of%20LLMs'%20Capability%20of%20Music%20Understanding%20and%20Generation%202024%20Accepted.pdf?sequence=2
  7. Why ChatGPT is the Best Music Recommendation Engine on the Planet
  8. https://openreview.net/notes/edits/attachment?id=oq5gUgJwkv&name=pdf
  9. 5 Expert AI Techniques for Unbiased Music Feedback to Elevate Your Sound
  10. https://openreview.net/attachment?id=QyLYF7SOpS&name=pdf
 
J

JustCKing

Superstar
Joined
Jun 17, 2012
Messages
26,037
Reputation
4,150
Daps
49,697
Reppin
NULL
I am not shocked at all about "Allure" being ranked so low. When the album dropped/leaked, I was in love with that song. Write a poem called "Allure" for British Lit my senior year and got an A.

Today, if I'm listening to TBA, its not a go to song for me. I rarely listen to the album as is. "I Know" which is also Jay Z/The Neptunes is superior. The drums kill "Allure" for me. Jay doesn't sound effortless on it, sounds like he's chasing the beat. The airy strings, the piano, and the little thunder effects are all dope. I would like to hear what it sounds like with the drums removed or replaced with a subtle bass drum without a snare or a muted, subdued snare.
 
You must log in or register to reply here.
Top