AI Amp Modeling Next Gen: A Frankenstein Scenario

[beans-on-toast]

Very little of public Github is public domain. Public domain means that no ownership is asserted and any use is permissible. Most code on Github is free and open-source code, but with formal licenses that have rules stating what is and is not permissible. Often Apache or MIT, which are liberal in what users are permitted to do with the code but do not allow everything possible, and they still assert clear ownership over the source.

This turns out to be a major problem when using an LLM to generate code: If it's appropriating source code which has been licensed by the Gnu Public License (another of the more popular FOSS licenses, and possibly the most famous, at least within software development), the terms of the GPL encompass all the code that incorporates the GPL-licensed code, meaning that your software is now also GPL-licensed. Which is of no meaningful consequence when you're hacking up a shell script to do a convenient file management task on your personal computer, but it has severe consequences in a corporate environment.

At my last coding job, before ChatGPT existed, we had to have license reviews before every software release, to ensure that we had not incorporated any code that had a GPL or non-free commercial license. Senior developers and lawyers were involved. It was slow, annoying, and had hundreds of millions of dollars at stake (as one of those companies making huge corporate backend software, we were less bothered with the possibility of our code being made public than we were with our clients suddenly being able to use our work without paying us any more). ChatGPT does not cite its sources, so relying on it can be a bad idea with massive consequences.

Companies have been using private code repository services hosted by Github and its competitors for over a decade. It's a common and, as such things go, reasonably successful form of cloud services, and had justified its existence long before ChatGPT came into being. Github, which is owned by Microsoft, which also uses Github as their own private repo, should presumably know better than to allow LLMs to scrape private code, even if Microsoft is perfectly happy to allow LLMs free reign over all the public repos it hosts.

ChatGPT can learn from public Github code without using it in a design. AI can learn how you want layout, functions to look, order of the functions, etc from public domain code to write in the style of a person. And yes, there are also private Github areas that are not available to the public.

Security is a bag concern, but it’s common these days for companies to use Amazon or other file servers as well as things such as Amazon Workspace for consultants that they hire. That way consultants don’t have accounts on company systems.

Licensing must be resolved as you said. It is a can of worms that will be more easily resolved with time I hope. Code generated should also be partially attributed to the AI system that helped with the design. Easier to deal with if you are writing open source code. Some companies make their code open source so that you’ll buy their hardware. Microcontrollers are one example, you want open source libraries.

I’m still not comfortable with cloud based designs.

 

[ardgedee]

ChatGPT can learn from public Github code without using it in a design.

True to a limited extent. The problem is that its code results can be a mixture of sourced code and its own logic glue. Github's own Copilot LLM is known to incorporate recognizable GPL code in its output. I haven't been able to find an analogous news item about ChatGPT's output but I'm inclined to blame how overwhelmed Google, Bing et al are with SEO-juiced links I don't want to read every time I use 'ChatGPT' as a prompt.

 

[agedhorse]

Ultimately, honest people, companies, countries will remain honest and dishonest ones will remain dishonest. That’s the way it’s always been.

 

[2thAche]

I think the bigger issue is the reason players play a bass… for the challenge and personal creative reward.

Would enough players accept the new technology, and would it provide the benefits that justify the considerable cost?

It wouldn't be players using this. It would be "creators" needing music for commercials, shows etc. AI is already replacing humans because it's faster and cheaper than people in many situations, and that's going to continue. I was just talking to someone who lectures / advises companies / develops AI and it's more prevalent already than I knew. The most interesting part is that when companies ask how the AI got to it's conclusion, or process it went through to generate data, they don't even know! All they can do is make models that this is *probably* how it got there. It's a bit scary but it's gonna be a continued reality.

 

[AgingEngineer]

It wouldn't be players using this. It would be "creators" needing music for commercials, shows etc. AI is already replacing humans because it's faster and cheaper than people in many situations, and that's going to continue. I was just talking to someone who lectures / advises companies / develops AI and it's more prevalent already than I knew. The most interesting part is that when companies ask how the AI got to it's conclusion, or process it went through to generate data, they don't even know! All they can do is make models that this is *probably* how it got there. It's a bit scary but it's gonna be a continued reality.

Mark my words: and "AI Psychologist" will be a job type in the future.

When a trained AI goes rogue (and they do all the time), there is no step-by-step instructions it follows. It has a model it's trained on, and recent experiences can change how the model us used to solve a problem. So like a kid that got their hand burned on a stove, and AI can ingest new data that will lead it to different conclusions.

 

[HolmeBass]

Wouldn’t need much if there’s a data connection, would it? Just Wi-Fi and a login to BassGPT premium, lol.

No way the network can support that in real time.

But running the model locally, maybe in task-specific hardware, could definitely work.

 

[InhumanResource]

No way the network can support that in real time.

But running the model locally, maybe in task-specific hardware, could definitely work.

It was really just an opportunity to say “BassGPT”

 

[jdjk7]

Exactly what I think, but we're old chaps, we won't last long. For the new generation, I think music is becoming just entertainment, it can be AI generated, they won't tell the difference. Or maybe there will always be a small market for real art and real music, and a large one for sheer entertainment. Hopefully.

Oh, I wish I was old*. I'm 26.

* No I don't

 

[-Asdfgh-]

So, do I lose credibility if I 'fess up to the fact that I loved the Monkees as a kid, and may (allegedly) still have a few of their tunes on a playlist?

Don't get up. I'll show myself the door.

I'd have to confess the same.

 

[-Asdfgh-]

Post-Human Music For Post-Humans.

I have a couple of friends who specialise in that. I like it, but it's .... challenging.

 

[-Asdfgh-]

This pretty much already happened, long ago, and it's had a pretty dramatic impact on music and the music industry IMHO. The resulting music tends to be sterile and lacking depth, nuance, and complexity. The songs would sound way more interesting if played by good musicians.

One problem is less and less people are actually learning an instrument, so there are fewer good musicians to play. Also since automation is much cheaper than paying real players, there is no longer enough money in the market to attract skilled musicians, or motivate them to develop the necessary skills. It's only going to get worse. But most people in the future won't know any better.

My impression is this problem is behind the current Hollywood strike. Unfortunately it may be a lost cause. Over the last year there has been a huge transition from scripted shows to reality BS and game shows. I believe the majority of scripted shows have been cancelled, most likely due to the price of producing them. The cancellations were announced before the strike. So what leverage do the strikers have to guarantee their future?

I can't speak for everyone, but this week the only show I will watch is the final to American Ninja Warrior which ran last night. I have watched this show for several years because it was the best thing on in it's time slot...good enough to pass an hour. As of this Monday, IMHO it was the best thing on for the entire week, which is pathetic.

Being played by bad musicians can be interesting.

I can't afford to hire a string quartet apart from at my wedding, let alone an orchestra to add to a song. That's why the Mellotron was popular. Plus I've never played on a stage big enough for an orchestra and a rock and/or roll band. Sometimes there has been enough for just rock or roll, no orchestra at all

 

[-Asdfgh-]

What you want is AI to generate songs that have a predictable chance of being a hit, or greatly admired by musicians. That you could then copyright, and thus convince a popular artist to create a cover. Some people are using the LLM AI's to generate songs. The question of AI work having a copyright is a legal issue that needs to be worked out. I don't think anybody would be able to tell AI from a Person creating a song. There's been enough drugs and alcohol involved in writing songs that AI will not likely be detectable as non-human

But since this thread is in an Amplifier forum - "Neural Amp Modeler" is a great way to experience what machine learning can do for sound:
https://www.neuralampmodeler.com/
An Amplifier Sound cannot be copyrighted.

Courts are indicating that AI-generated compositions are not eligible for copyright.

 

[-Asdfgh-]

I'd have to confess the same.

It was top notch pop music of the period, or even today.

 

[N. Kentucky Bass]

It was top notch pop music of the period, or even today.

Yes, and ironically, the Monkees didn't even play the music on their records. LA session pros did that for them. So the B.S. never changes.

 

[seamonkey]

Courts are indicating that AI-generated compositions are not eligible for copyright.

Yeah, anything not created by a human can't get a copyright, but if a human uses AI as a tool to create a composition, the human can get a copyright based on their contribution. I don't think anybody is going to be able to tell what the AI contribution to the composition was greater than the human.

 

[ardgedee]

Yeah, anything not created by a human can't get a copyright...

That's true within the US (at least for now), but not necessarily settled law in other countries.

 

[larrysb]

Yes, and ironically, the Monkees didn't even play the music on their records. LA session pros did that for them. So the B.S. never changes.

I wouldn't be so quick to condemn the Monkees. I mean it was actually a TV show, they had a tight production schedule to keep up with. I mean they basically had to turn out 2 to 3 "hit-quality" songs for every week's episode. That's why you hire out the best session talent in LA, they're fast, they're good and they're pretty expensive. That practice continues through to today even

Nesmith was a credited writer and composer as well as a capable musician. The rest of the "Monkees" cast were all legitimate and capable musicians and vocalists. Jones actually sang the tracks. There ya go.

Being a TV show, they were brought together for their photogenic and acting capabilities.

How many records have Carol Kaye and Leland Sklar been tracked for?

What makes for a great touring band, doesn't always translate to a great studio band. Nothing wrong with hired guns in the studio in my opinion.

 

[-Asdfgh-]

Yeah, anything not created by a human can't get a copyright, but if a human uses AI as a tool to create a composition, the human can get a copyright based on their contribution. I don't think anybody is going to be able to tell what the AI contribution to the composition was greater than the human.

I'd say that the extent to which even use of a AI as a tool might affect copyright still needs to be tested in court. It's already been ruled that you can't simply have an AI write all possible melodies over a weekend and claim that your human input is selecting all of them and copyrighting them all. So simple selection may not be enough but what sort of additions, assemblage or whatever is required remains to be seen. These are uncharted waters and the law on the use of randomness or algorithmic composition is mixed

 

[-Asdfgh-]

I wouldn't be so quick to condemn the Monkees. I mean it was actually a TV show, they had a tight production schedule to keep up with. I mean they basically had to turn out 2 to 3 "hit-quality" songs for every week's episode. That's why you hire out the best session talent in LA, they're fast, they're good and they're pretty expensive. That practice continues through to today even

Nesmith was a credited writer and composer as well as a capable musician. The rest of the "Monkees" cast were all legitimate and capable musicians and vocalists. Jones actually sang the tracks. There ya go.

Being a TV show, they were brought together for their photogenic and acting capabilities.

How many records have Carol Kaye and Leland Sklar been tracked for?

What makes for a great touring band, doesn't always translate to a great studio band. Nothing wrong with hired guns in the studio in my opinion.

And Dolenz was an innovator in electronic music, although largely ignored because he had been in The Monkees.

 

[-Asdfgh-]

Yes, and ironically, the Monkees didn't even play the music on their records. LA session pros did that for them. So the B.S. never changes.

They played on later albums. They wanted to play on them from the start