Thoughts on AI Analysis
AI-driven analysis comes with all of the positives and negatives of other computer-based styles of analysis. It has the capacity to generate and sort enormous quantities of data at speeds alien to human researchers. It can also, however, incorporate errors, inconsistencies, and irrelevant information seamlessly into these datasets, forcing human researchers to check the computer’s work or even toss out the data entirely.
What appears to be a new development in AI’s case is its unique ability to source data from third-party sources. This introduces a new set of complexities; this data cannot be fact-checked by human researchers if they are unsure where the information was gathered. The larger the project, the easier it becomes for such problems to slip into the finished product. As such, I am cautious to speak positively of their research potential at this time though, as I have mentioned at previous points, the potential for an exclusively academic-run AI is enticing.
Project Proposal
Searching through the Cultural Analytics Lab’s pages, I came across an article titled “Seeing a Century Through the Lens of Sovetskoe Foto,” by Alise Tifentale of the City University of New York. This project, carried out alongside Lev Manovich and AgustÃn Indaco, analyzed Soviet photography based on the covers of the USSR’s state-sponsored photography magazine over the years it was published.
I suspect that a similar project could be applied (with considerable additional effort) to the material culture of far older groups. I (cautiously) suggest that a database of surviving artifacts from Antiquity could be formed in three dimensions, allowing AI to search for comparisons across cultural groups. With appropriate safeguards in place for false positives, this could become a particularly useful tool for trade-heavy periods and regions in which similar items in different cultural styles frequently crossed borders. The AI could potentially identify cultural exchange and the drift of artistic style.