There couldn't be a better time to be analytical about the usage of AI in general, particularly in architectural conceptualization and visualization. Over the past two years, I have been investigating and making predictions, observing how AI is being integrated into various fields with the development of large models, social media, and popular views.
From my perspective, the overall quality of these models' outputs is in an exponential decline. Whether in terms of realism or aesthetic innovation, the outputs seem to be deteriorating. Another aspect is the consistent connection between these models' outputs (mainly images/videos) and social media. This results in outputs that follow current trends, influenced by biases inherent in both the users and the tools.Recently, there has been a persistent urge to incorporate these tools into current practices rather than exploring different approaches to revolutionize the field or at least navigate through a practice that is deeply deceptive from both inside and outside.
For instance, in this example with MidJourney V6.1, which essentially references different versions of MidJourney V5 and V6, the newer model has failed (or maybe it's me) to replicate anything interesting from the older models. It's as if it has become excessively biased, resulting in photorealistic outputs that lack the innovative qualities I found compelling before.
As we (or perhaps more accurately, I) continue to endeavor to understand the capabilities of these tools, the future of their development and usage remains uncertain. This uncertainty is typical of any new technology, characterized by huge potential heavily influenced by capital contexts, the hidden conformity of the post-digital era's individuality, and a view of the future that is completely divorced from present reality.
Komentáře