Categories
Updates

Music Produced By AI? Will AI Soon Replace Artists? Probably.

Midjourney’s interpretation of an AI creating music

It’s on everyone’s mind – what’s AI going to do to music?

The opinion pieces out there tend to be starkly Manichean in their outlook. AI will either empower creators in ways never before imagined or destroy those professions wholesale.

At present, both of those viewpoints are true to a certain extent, but one can’t help feel the shadow of the latter looming large over most discussions.

Writers and illustrators have already had their professions impacted in a significant way with tools such as ChatGPT and Midjourney becoming as prevalent and mainstream as common household appliances. 

The big question of the day is what impact AI will have on the music industry – and specifically on artists and musicians.

Currently, these folks share many of the same concerns about AI that are fueling the current Writer’s Guild of America (WGA) strike (and which was recently joined by members of SAG-AFTRA, the union representing over 160,000 film and television actors, among others. Namely, that AI is threatening to encroach into intellectual property and creative control in ways that disenfranchise those creating the actual music we love and enjoy.

These fears aren’t hypothetical either. Unlike the television industry where there hasn’t (as far as we know) been any well-received or rival productions that were entirely AI generated, the music industry has already had a couple.

Back in April, an AI-generated track that mimicked the vocals of Drake and the Weeknd went viral and was found on multiple streaming services such as YouTube and Tiktok before being taken down. Despite

Hell, just last month, there was news that a “new” Beatles song would soon be released using AI. Based off a 1978 demo using John Lennon’s voice, the AI track was spearheaded by one of two surviving Beatles’ members Paul McCartney.

So, will AI be replacing our pop stars and starlets in the near-future? In this case, I would go with most likely, and there’s a few reasons why.

1) Some music is already being replaced by AI and likely won’t go back to being produced by musicians.

If you’ve ever been put on hold during a call and heard a jingle or noticed a certain non-descript pop track in the background of an audio ad, that was likely stock music available from one repository or another.

Historically, stock music has been its own thing separate from the kind of tunes we hear being released by record labels. Most stock music is produced by musicians either creating tracks on commission, as freelancers, or just releasing tracks with a copyright that allows it to be used for commercial purposes.

While these compositions tended to be simple arrangements, often only a minute or two in length, they’ve still historically needed someone to handle the composing, arrangement, recording, and all the rest.

Nowadays, that’s already changing. While most stock music might not currently be AI produced, it almost certainly will in the next year or so.

One reason for this is because there’s no shortage of AI music generators available on the web. Sites such as Boomy, Soundful, and Soundraw, among others, offer users the tools to generate unique AI produced tracks in a variety of genres.

To be honest, some of what is produced by these isn’t half bad and doesn’t sound all that different from royalty free stock music that’s already available out there. The main difference is that where an artist might need an afternoon to put together a stock music track the traditional way, with AI it takes seconds – and each piece is unique.

And so, while you can still go to a stock music site and browse what’s available by genre, length, and all that, paying a subscription fee as is usually the case, a producer could also simply sign up for an app and create it all themselves – cutting out the middlemen.

It’s not hard to see the appeal in that, especially as it cuts down the cost and time for finding the right track. However, it does mean that the folks who have traditionally been producing royalty-free stock music are going to be facing stiff competition from AI tools and tracks alike.

2) AI is already able to produce backing tracks and instrumentation on command.

Similarly, to how we have tools to produce stock music, there’s little in the way from using those tools to create tracks that accompany an artist’s lyrics or just having someone sing over tracks spun from an AI tool.

However, the real deal is the ways in which AI plugins and tools are becoming available for the tools that producers are already using to put together their tracks.

If you do a quick search for AI tools available with FL Studios (formerly known as Fruity Loops), you’ll come across a couple of dozen plugins already available – not to mention, a whole ton of YouTube videos giving tutorials for how to blend AI into your productions.

The shortcuts these tools offer is pretty obvious – no more writing out a catchy baseline or piano to accompany your chord progression. In a couple of clicks you can have an AI churn out a few options and pick the one that goes best with what you had in mind.

As someone who creates music electronically (I’ve been using Reason on and off for the past 20 years to self-produce electronic tracks), it’s both a little disheartening and somehow empowering at the same time.

On the one hand, all the work that goes into the discovery process of putting a track together can be automated with little blood, sweat, and tears.

On the other hand, as someone who also has less free time than they used to, the idea of generating the basics with the click of the button is an appealing (if also kinda frightening) shortcut.

In any event, if tools like this exist, people will use them.

3) AI generated music is starting to flood streaming services.

Up and coming music producers, indie artists, and DIY folks like myself, have been using platforms like Soundcloud, YouTube, and even MySpace back in the day, to release music, give it a home, and hopefully find an audience.

It’s tough enough to get any attention even in a niche genre as at any point there’s hundreds, if not thousands, of others also releasing their own tracks.

AI generated tracks are going to create pure pandemonium on these platforms.

Giving anyone the ability to generate a track in under a minute means that on any given day, we’re likely going to be seeing the number of new releases skyrocket.

Sure, AI most AI tools are generally urging users to keep this music for their own projects, but there’s little to stop someone from labelling it something they self-produced and then start promoting it far and wide.

Worse, while the current tools are a little predictable in their output, and hardly as creative as even your average suburban garage band, because AI is driven by machine learning – they will learn and become more advanced, and likely more quickly than we realize.

I wouldn’t be surprised if by the end of the year, half the streaming platforms become flooded with AI compositions not clearly marked as such – and whether we will be able to even tell the difference will be up for debate.

4) The deepfakes are real, and only going to become more common.

Much like how the recent Black Mirror episode “Jane is Awful” played with the notion that actors might soon be giving up rights to their voice and image (mirroring an actual concern of SAG-AFTRA in their strike), it’s hardly a remote possibility that musicians will face the same concern.

We know that AI can already be trained to mimic an artist by listening to their recordings and be used to create convincing recordings (to circle back the case of Drake, The Weeknd, and John Lennon). So far it’s mostly been either industrial folks on the internet creating deepfakes, or surviving bandmates making the call, but how long before studios begin deciding when AI gets involved?

Let’s not forget that a few years back, 2Pac was “resurrected” as a 3D projection some time ago. It won’t require leaps and bounds to make a more autonomous 3D rendering of an artist start to compose new songs.

I strongly suspect it’s only a matter of time before contracts with record companies also start including clauses about who controls ownership not just of an artist’s current discography, but also their likeness and vocal styles.

When (and not if) we get to that point, will the listening public even care if a catchy song was produced by the original artist of an AI bot?

And even if there is an initial outcry, will it last or (like most things) will anybody even care?

Categories
Updates

Movies written by AI? Will AI Replace Screenwriters? Maybe.

Midjourney’s interpretation of ChatGPT writing a screenplay

If news about Hollywood and screenwriting passes your inbox on a regular basis, you probably heard about the Writer’s Guild of America (WGA) strike that’s been going on since early May of this year.

While this is not the first such strike to happen in our generation (there was another one back in 1988, and more recently in 2007), it’s certainly the most unique of the three where its demands are concerned.

Some of the demands are standard grievances – a desire for greater compensation for writers, particularly when the residuals from streamers are involved – but the big sticking point this time around seems to be about something else: AI.

Why are writers from the WGA striking about AI?

In March of this year, OpenAI (the folks behind ChatGPT), published a study looking at potential labour market impacts with the adoption of AI. Writers were placed in the category of most likely to be impacted.

In other words, AI will not only greatly impact the way writers do their work but also whether they continue to do their work at all.

These developments haven’t gone unnoticed by the WGA.

It’s not that everyone is worried that screenwriters will be wholesale replaced by AI either (at least not this week).

In it’s current state, Ai cannot produce a workable, original, or even engaging screenplay. Sure, if you prompt it correctly, it can produce something that looks like it, but with about as much life and inventiveness as something a fifth grader might produce for a school play.

Rather, the real sticking point is that writers want to have more say and agency in deciding how and when AI is used in the writing process.

Specifically, the WGA wants a claus in their contract that stipulates that no writer will be required to adapt anything that been output by AI. In other words, that they cannot be contractually obligated to complete or revise some partial screenplay produced by an AI.

What does this mean in practical terms?

Well, take for one that it’s not uncommon for screenwriters to get called in as either ghostwriters or as script doctors to polish up an already or mostly finished piece before it goes to production.

It’s not inconceivable that some unscrupulous studios might try to write the proverbial Shakespeare play out of a room of monkeys – only in this case the monkeys are ChatGPT.

It’s worth noting, therefore, that the WGA isn’t saying they refuse to work with any AI in the writing process – but rather, that the screenwriter can continue to control the production process of their work. Sometimes that might involve using an AI to clean a phrase, find synonyms, or just generate ideas – but it will be the writer’s discretion, and not the studios to decide in what capacity.

How much of a threat is AI to writers?

When this round of negotiations first began, there were clauses about AI brought to the table. However, because of how rapidly these tools advance, the initial concerns seem like the distance past – and because of this, it’s most likely going to be difficult to nail something down on writing that is future proof.

And speaking of the future, how much time do we have before AI has advanced to the point of either doing the bulk of the work or becoming a necessity in any creative process?

Well, it’s hard to say.

A quick Google search for “can AI replace screenwriters” turned up about a million and a half results, showing that it’s a topic a lot of people are wondering (and worrying about).

However, the results are anything but in agreement.

The first place result was from Wired UK, titled “Hollywood’s Screenwriters Are Right to Fear AI“.

Immediately after it, was a piece from TechCrunch titled “AI can’t replace human writers.”

If you only read one headline, you might walk away convinced of one side or the other – but the presence of all these differing viewpoints (not to mention Reddit debates) suggests its very much uncertain.

So what are the facts on AI at this point in time?

  • AI cannot at present write a workable screenplay (except maybe as a curiosity … I wouldn’t be surprised if someone produces an AI screenplay as a sort of art house experiment).
  • When used as a tool, AI can cut down on the work and hours that go into producing a screenplay.
  • It’s also not clear who owns the rights to a script produced by an AI. If ChatGPT turns some prompt into a workable script, would that mean ChatGPT and its owners receive credit and ownership of it when it goes into production?

This might look like screenwriters are safe for the time being, but I suspect looks are deceiving.

Technology advances more rapidly than we suspect. Just look at the image generating apps like Midjourney. A year ago, it took sometimes a dozen rolls to get a workable image. Now, it often produces exactly what you need on the first go.

When ChatGPT and Bard become more fully integrated with the internet and live content, they will only learn so much faster. The time when an AI can produce a workable script might be closer than when we think.

And let’s not forget that studio heads – and the ones holding the wallets for the film and movie industries – are also pushing for this. It’s no secret that most writers are already undervalued for their time and contributions. If writing rooms across the room are already staffed by underpaid and overworked folks, how much greater will their woes be when AI starts sitting at the table?

As a writer myself, I’m not expecting to be competition free any time soon. The web is already filled with bots writing blogs, answering threads on socia media, and producing content that people read daily. I can’t help but feel it’s only a matter of time before we start reading books and watching TV that’s largely produced by AI.

And in case you’re curious what that could be like, an entirely AI driven show already exists on Twitch called Nothing, Forever, as a sort of insanity take on Seinfeld.

Categories
Updates

Upcoming Horror Publication: “Magpie” in Under the Stairs

My psychological / paranormal horror story “Magpie” about a couple moving back to a family home with possibly a few ghosts of its own, was recently accepted by Under the Stairs press.

Written under my much easier to spell (and pronounce) pen name JM Connors, the story will be appearing in anthology along a dozen or so other stories later this year (September 2023).

I’ll update this post when I have more information about the release.