The advent of AI is rapidly revolutionizing the field of virtual archaeological reconstructions, which have hitherto been carried out using settings with costumed sets and extras or 3D technology. 3D technology, especially when integrated on the photogrammetric reproduction of the archaeological site, offers greater scientific rigor. But the rapid evolution of generative artificial intelligence already offers a wide palette of possibilities, with surprising results. The challenge is to bring it back to the necessary rigorous reference to scientific sources. Archaeological reconstructions refer to many types of sources. Of considerable importance are physical finds that can identify the historical period and architectural styles, ancient maps or views or cartographic surveys regarding architectural and urban layout, but also details on decorations, furnishings, and finally textual descriptions from ancient authors and modern scholars. Any historical reconstruction, regardless of the technology to be used, requires a preliminary collection of scientific information. Once the necessary data has been collected, whether written descriptions or archaeological traces, moving on to reconstruct the complete building is, however, extremely difficult, especially if the building is to be reconstructed realistically, rich in details, decorations, and even in color. Remains still found at archaeological sites, literary descriptions, and schematic drawings on ancient coins or in bas-reliefs shed light on some aspects of ancient monuments, but they are far from offering a realistic picture of them. The data at our disposal are always insufficient, and if one wants to obtain a “photographic” rendering of the virtual building, it is necessary to go beyond the limits of demonstrability and boldly venture into the field of hypotheses. It should be pointed out immediately that archaeological hypotheses are sometimes conflicting, on the structure of the building, not to mention its decorations and colors. or therefore exposed to the possibility of erring, rather than omitting the chromatic, decorative and figurative elements that undoubtedly characterized the ancient monuments. Once all informational resources have been used and thus exhausted, it is necessary to proceed by being guided by historical intuition and aesthetic sense, which must guide historical regeneration in detail. Science and art must guide artificial intelligence to make the past come back to life, becoming a rigorous and consistent synthetic reality. Compared to 3D reconstructions Generative AI seems to provide more vitality and complexity, but it requires more imaginative effort and more attention to detail. Thanks to AI, it will be possible to shape environments that do not reproduce bare architectural forms, but recreate the ancient symbiosis of architecture, painting and sculpture, experienced by hyper-realistic human figures. Generative artificial intelligence offers an extraordinary opportunity to rethink the ancient world as a whole, offering a new image of it to the general public, bursting with figurative and symbolic values and enriched by realistic human presences, which were difficult to achieve with previous technologies.
Italiano