> But global paradigm shifts aren’t pleasurable for everyone. As I explained in my latest article on AI art, “How Today's AI Art Debate Will Shape the Creative Landscape of the 21st Century,” we’re getting into a situation—now accelerated with the open-source nature of the model—that’s extremely complex. Artists and other creative professionals are raising concerns and not without reason. Many will lose their jobs, unable to compete with the new apps. Companies like OpenAI, Midjourney, and Stability.ai, although superpowered by the work of many creative workers, haven’t retributed them in any way. And AI users are standing on their shoulders, but without asking for permission first.
As I argued there, AI art models like Stable Diffusion pertain to a new category of tools and should be understood with new frameworks of thought adapted to the new realities we’re living in. We can’t simply make analogies or parallelisms with other epochs and expect to be able to explain or predict what it’s going to happen accurately. Some things will be similar and others won’t. We have to treat this impending future as uncharted territory.
I wonder if we'll also talk about "conversations", "complex situations" and "the need to treat this as uncharted territory" when some Copilot/GPT3 successor a few years down the line spits out entire production-ready software stacks off the prompt "like Facebook only better" - using our own code as training data.
This prompt is unspecific to the point of unusableness. Even if this works some day, the spec used will be a lot more detailed, in higher-level pseudocode style.
True, but as you can see with image generators, they can without any problems work off extremely underspecified prompts, they will just use their own priors to fill the gaps.
There will absolutely be prompt engineering and I agree that actual, serious prompts will be much more specific than that.
I don't think the prompts will necessarily be pseudocode-style. Depending on what trainsets are available, I could imagine we'll have some high-level description of desired features in addition to lots of specifiers which narrow down the specific languages, design patterns, tools etc which should be used in the resulting codebase.
You can already use similar prompts with Copilot today by disguising them as comments.
What you're describing is essentially a DSL. We can ruminate on the exact level of details that will be needed but in my opinion by necessity it will be a lot higher than what most commenters in those threads are imagining. Open API specs are already quite complex and verbose and it's not exactly clear to me we can do a lot better when describing APIs, whether web ones or not.
As I argued there, AI art models like Stable Diffusion pertain to a new category of tools and should be understood with new frameworks of thought adapted to the new realities we’re living in. We can’t simply make analogies or parallelisms with other epochs and expect to be able to explain or predict what it’s going to happen accurately. Some things will be similar and others won’t. We have to treat this impending future as uncharted territory.
I wonder if we'll also talk about "conversations", "complex situations" and "the need to treat this as uncharted territory" when some Copilot/GPT3 successor a few years down the line spits out entire production-ready software stacks off the prompt "like Facebook only better" - using our own code as training data.