[torben-friis]

Apple has a potentially interesting use case for generative AI in their professional creative apps: heavy integration in logic pro or in final cut. Perhaps even create simpler tools with similar functionality but aimed at non professional users.

The problem is that this risks antagonising the everyone in arts/humanities, and most other use cases are really unneeded - who needs text summarizing for something as simple as personal texts from friends? casual use is not really complex enough to warrant an assistance.

[xena]

Author of the article here. I do video work occasionally and I use Davinci Resolve to do it. Davinci resolve uses generative AI as tools to help you. It makes all my subtitles and if I'm not going into domain specific terminology that often, it'll be 95% of the way there in about 15 minutes. This is massive, especially when combined with "edit by word" editing.

[CharlesW]

FWIW: Speech-to-text falls under "AI", but is not considered generative AI. (Note that systems with capabilities that go beyond STT with capabilities such as summaries or translation may incorporate generative AI.)

[ChadNauseam]

> The problem is that this risks antagonising the everyone in arts/humanities

I don’t anticipate this being a problem. Have you used generative fill in photoshop or lightroom? It’s a complete game changer. In Egyptian mythology they weigh your soul against a feather when you entered the afterlife, and with professional tools I think moral hangups about AI are going to get about the same weight. It’s just too good not to use.

[duped]

I have this deep feeling that engineers have a fundamental misunderstanding of the arts, which is reinforced when there is a suggestion that "heavy integration" of generative AI into multimedia production apps is somehow desirable. It's not just contrary to the design and use of these applications, but contrary to art as an endeavor - and users find it revolting.

Apple already has simpler tools aimed at non professionals, they don't need generative AI either.

[torben-friis]

>It's not just contrary to the design and use of these applications, but contrary to art as an endeavor - and users find it revolting.

As far as speaking purely about art goes, I think there is a wide debate to be had there - a ruler helping a line be straight is help to an artist but not seen as contrary to his work, while pressing a button and getting a full painting is clearly not art creation. But where in the middle lies the spot where automation stops being ok? I think it's a spectrum and we'll see a shift in perception there, gradually.

But that debate completely sidesteps the elephant in the room - most artists nowadays don't make a living making art, just making art-adjacent content, where the artistic value is not really super appreciated by the buyer - photographers creating stock photos, graphic designers making app icons, background music for ads and the like.

Artists hate tools that automate this process because it significantly removes that source of income, but they're not the main target of these products. The target is the clients currently paying them and seeing an opportunity to get a product that, while lacking artistic quality, works for them just as well.

[duped]

This is another place where I think technologists miss the forest for the trees. You're looking the outputs and results looking for a middle ground, but misunderstanding the problem of generative AI in art is the act of creation itself.

People don't generally take issue with tools that automate or make their jobs easier, even if it may reduce the value of the output. However if the tools limit what they can create themselves and make it difficult to fix or fine tune when something is not how they envision things in their mind before creating it, then they're not good tools. Even worse are the tools that take away their ability to create at all.

Really I think what technologists don't understand about art is that in engineering tools are a means to an end and only the outputs matter. If you can get a program to spit something out and say "look, isn't that good enough?" you have missed the entire point of art.

[torben-friis]

>However if the tools limit what they can create themselves and make it difficult to fix or fine tune when something is not how they envision things in their mind before creating it, then they're not good tools. Even worse are the tools that take away their ability to create at all.

I might be wrong, but I think you're picturing all-or-nothing use cases here. It's not all just 'draw me a picture'; Think smaller scope and maybe you see that middle ground. Take as an example, for a writer, clicking on a phrase like 'he raised his eyebrows' and being suggested alternative wordings so he can avoid repetition. Is that interfering with his act of creation any differently than checking a thesaurus?

Consider being able to have an interaction with an LLM to whom you can ask 'is the plot of my thriller so far leaving any plot hole?'. That does not seem so different with a back-and-forth with an editor or an early reader, in terms of affecting creative freedom.

>If you can get a program to spit something out and say "look, isn't that good enough?" you have missed the entire point of art.

Again, I get that but art is not what tech companies are trying to substitute. If a music generator can give you background music for studying there is no art creation involved, but neither the owner of the youtube channel making ad money nor the listeners give a shit.

I'm not defending that position necessarily, mind you, just pointing out that the business interests in 'not art, but just content, that happens to need artist's skills to create' far surpasses the interest in actual art.

As an analogy: Many musicians will scoff at mainstream pop artists and how every song is just the same four chords. But is the business in pop or in avant garde jazz?