- This topic has 12 replies, 5 voices, and was last updated 2 years, 2 months ago by
.
-
Posts
-
-
While I tried ChatGPT for the first time a few days ago, it is actually quite interesting. I wanted to re-write some text for a project. So after trying a few things decided to see if it could write some lyrics for a blues song. In my opinion it does a pretty good job, gives you some verses, chorus and a bridge. You can regenerate the lyrics if you don’t like them. Just give it a little hint in the prompt about the blues topic. Interested to know if anyone else has tried this and what they think about it. ChatGPT is actually very easy to use.
-
-
I haven’t used it, because I don’t write lyrics. But my grandkids are in high school, and they told me that it’s common for the high school rock bands to use ChatGPT to write lyrics. And pretty much all high school kids use it for writing essays, book reports, etc.
I don’t know if you’ve ever paid any attention to the lyrics of modern pop and hip hop songs, but I’ve seen quite a bit (like 70% of people now days, I stream TV with the subtitles on), and Chat GPT lyrics might actually be a huge improvement.
Sunjamr Steve
-
Steve, I don’t really like most of the modern pop music so agree chatGPT might be an improvement. Of course if you listen to some of the 60, 70, and 80’s music there are some bad lyrics.
I don’t normally write lyrics either, but thought it would be fun to take a chord progression and put some lyrics to it. Might give the music a little more feel. John Fogerty would sometimes make up nonsensical words when writing a song until he magically created the finished lyrics. -
No offense intended, but I can’t resist chiming-in to say that – while I value some level of technology, I resent the potential for takeover of ‘human-ness’ by AI and the like, especially regarding something so filled with human emotion, like music and crafting lyrics.
As one who dearly loves writing lyrics that come from the heart and soul, (sometimes they work, sometimes they don’t making me have to work harder on it, there-by offering potential for improvement!) I have not found the value in “artificial intelligence” (Operative word being “artificial”) writing love songs.
If “ANYBODY can do it” so easily without the touch of humanity, passion and craftsmanship where IS the artistic value?I hope the dialog and debate are taken in just that spirit and not taken as insult as it is not meant to insult, just adding my personal (74 yr. old) opinion to the conversation :o) I realize my generation and ‘old school’ ideas are living on a short calendar, but while I’m still here, I feel compelled to offer my two-cents worth. :o)
Regards,
Marie“Laissez les bon temps rouler“
Marie Jarreau -
-
I don’t know if you’ve ever paid any attention to the lyrics of modern pop and hip hop songs, but I’ve seen quite a bit (like 70% of people now days, I stream TV with the subtitles on), and Chat GPT lyrics might actually be a huge improvement.
On the cynical side, but I fully second that!
In the “good old days” (early 80’s) the AI field was trying very hard to reproduce the reasoning of subject matter experts, and eventually replace them. Remember the so-called “expert systems” of the days? The only way to program them was with the help of a field specialist (physician, geologist, chemist, …).
But with a few spectacular exceptions, it failed at larger scales because programmers realised that the key was not so much the “reasoning process” but rather the huge context that was inevitably needed to solve real world problems, and there was simply no way to provide that context information.With the advent of the internet and the ability to tap into that huge distributed data reservoir, the problem was actually approached from the other end. Instead of programming a reasoning process, just mine the data and “create” the solution automatically using a handful of relatively simple (but very effective) algorithms.
In a way, you could say that the right answer is produced just by looking (smartly) at the huge amount of information available about any particular problem. The more data available, the higher the probability to be within the statistical boundaries.That’s exactly what AI is today. Please note that any AI engine, ChatGPT included, needs to “trained”. That is to say, it needs to be told what’s correct and what’s not correct so it can refine its algorithms. (We have now almost reached the point where AI engines can train other AI engines).
But an AI engine only uses (can only use) human produced information to start with.
So when it “composes” the lyrics of a blues, it reuses human feelings and lyrics – it doesn’t invent them. And it certainly doesn’t feel them.I don’t believe that AI will ever surpass humanity in creativity and “replace” us human beings. IMO, the big danger of AI today is precisely that it potentially embarks and amplifies all the negative aspects of humanity: racism, inequalities of all kinds, greed, hate, etc., because all that is also abundantly available in the “big data” and if used improperly to “train” an AI engine, you will end up with a hell of a monster.
-
Jean-Michel,
I do agree on needed improvement for the modern pop and hip-hop lyrics. (RAP-lyrics is a completely other world!)
Though, even there, I chalk it up to “beauty is in the eye of the beholder” {they have had a following) with the caveat that “beauty” is also a human experience. LOL! I begrudge no one their personal ‘taste’ in music. :o<I also “get” your comments on the cynical side. AI WILL amplyfy the negative as well as any positive fed to it by the human ‘trainers.” Part of my concern is that while humans can make mistakes that can cause harm in programming due to their own beliefs and ideology, at least the human can undergo a change of heart and ideology. From what miniscule information I understand about AI entities/systems, the effort to then undo, redo or upgrade the AI “training” could, also be a monsterous undertaking – if there was such an effort. So once AI has proliferated its information – it’s “out there.” It could be available to influence society long after the programmer has recognized his biased shortcomings.
Anyway, I guess that Pandora’s box is already opened and we know what that means . . .
Thanks for the discussion.
Regards,
Marie“Laissez les bon temps rouler“
Marie Jarreau -
Marie,
No offense taken, I like to hear different points of view.
I just look at AI in its present state as a tool for saving time and maybe helping human inspiration along.
My daughter uses it at work and it saves her a lot of time on boring and routine things, giving her time for more important matters.
Eg. you can have AI design a physics test for a school exam, for either written or multiple choice (all done in couple of mins). Is that cheating? I don’t think so it’s not the real test you’re just doing your homework more efficiently.
In the big picture it is a little scary because the bad actors will use it but the genie is out of the bottle so hang on. -
Hi Gordon,
I think you’re right, and unfortunately the genie’s bottle was shattered during his last move and so he actually came out of Pandora’s box – we’re really up a creek!🤣Thanks for the dialog. Differing views bring us to better understanding of the world and ourselves.👍🏻
Good Wishes,
Marie“Laissez les bon temps rouler“
Marie Jarreau -
-
-
I had a thought. Humans influence one another, right? In order to achieve that influence they have to be, in the minds of the influenced humans, imbued with a certain amount of trust. This is the primary reason I do not trust AI in its present state. There is no transparency, no reputable independent consensus, and no laws to enforce accountability.
Maybe the key to this is AI developers need to emulate how humans influence others, and why it works. This means imbuing their systems with higher level of trustworthiness. If they can’t demonstrate that they deserve some degree of trust, and their competitors do, they will be at a disadvantage in the marketplace.
If we want to further emulate human influence we have to factor in that humans basically split into overlapping zones, like a Venn diagram. A political movement attracts others who confirm some of their preexisting biases. Logically, the more radical the AI developers’ stance, the fewer other zones they will overlap, and the fewer people they will be able to influence. They will drift out onto the ‘lunatic fringe’ in other words. Does this sound familiar? 🙂
If this sounds like a centrist position on my part I guess it is. Maybe AI evolution will recognize this and become more or less self correcting. Actually the more accurate term would be ‘self moderating’. I think that’s a very long way down the track though.
Just a thought.
-
- You must be logged in to reply to this topic.
