Google launches an online retrospective for South African artist Dr. Esther Mahlangu, using AI to recreate early memories

Google is rolling out a new online retrospective for South African artist Dr. Esther Mahlangu on Google Arts & Culture, positioning the project as both a cultural archive and a test of how artificial intelligence can help preserve personal history. The initiative, introduced in a Google company blog post dated November 11, is being released in recognition of Mahlangu’s 91st birthday and centers on a first-of-its-kind effort to fill gaps in documented material from her early life.

The exhibition is built around more than 45 curated stories and a large archival collection that, according to Google, spans Mahlangu’s art, life, and legacy. It also includes content designed for remote viewing, including virtual tours and references to her murals and Ndebele cultural traditions, which the company says can now be explored from anywhere in the world.

A key feature is the handling of missing early-history records. Google said that while it assembled research materials covering Mahlangu’s work, the archives did not include items from her life before 1980, leaving “crucial pieces of her history” absent. To address that, the project team used AI to recreate parts of her childhood memories based on her own stories, aiming to preserve heritage while acknowledging that the materials were not otherwise documented in the archive.

Google described the collaboration behind the retrospective as an effort that brought together curatorial and archival work with technical expertise across Google DeepMind, Google Arts & Culture Lab, and a synthetic memories initiative led through Domestic Data Streamers. The team said the work was developed through a series of workshops and interviews that focused on documenting a missing period, from when Mahlangu began painting at age 10 until her first documented works in 1980.

In describing how the AI component was formed, Google said the process began with still images generated from Mahlangu’s memories, drawing on the recollections and the available archival sources. It then used Veo, Google’s tool for turning images into videos, to animate those early recollections into videos. Google cited one example from those memories: Mahlangu’s account of discovering her talent for painting as a 10-year-old, using feathers instead of a brush.

The retrospective also draws on multiple partners that Google said collaborated on the curation, including Art of Contemporary Africa, Iziko Museums of South Africa, the British Museum, the Jean Pigozzi Collection, the National Museum of Women in the Arts, and Serpentine. Google further said the project incorporates context for how Mahlangu became known globally through her artwork, archival material, Street View, and other media elements used by Arts & Culture to connect audiences with sites and works.

Google framed the effort as more than a one-time publication, arguing it demonstrates “a new way to approach the preservation of cultural heritage” while keeping “respect for personal memory and historical accuracy” at the center of the process. In the blog post, curator Nontobeko Ntombela, who is identified as a Senior Lecturer at the University of the Witwatersrand’s School of Arts, said her aim was to present a comprehensive exploration of Mahlangu’s life and work and that the retrospective was shaped by an ongoing research relationship with Mahlangu and her family.

From a business and technology standpoint, the retrospective reflects how major platforms are increasingly using AI not just for search and recommendations, but for cultural presentation and storytelling. For Alphabet’s Google, Arts & Culture is one of the more visible outlets for that approach, and the introduction of Veo-style video generation and AI-based reconstructions indicates that the company views synthetic media as a way to extend archives when documentary gaps exist.

Still, the project also raises questions that are not fully resolved in Google’s announcement. The blog post describes the intent and the workflow for generating visuals from memories, but it does not spell out how the team validated accuracy beyond the use of Mahlangu’s recollections and available archival materials, nor does it provide a clear breakdown of which specific childhood events were reconstructed and what level of uncertainty remains for viewers. As with other uses of AI in content creation, the line between interpretation and documentation may matter to audiences seeking historical evidence.

For viewers and for the arts-tech industry, the next phase to watch is how Google and its museum partners present provenance and context around AI-generated reconstructions within the Arts & Culture exhibit itself. The company’s claim that the work is guided by respect for memory and historical accuracy suggests that future cultural projects may increasingly pair synthetic media with curatorial explanation, and the reception of this launch may influence how widely that model spreads. Given the emphasis on preserving heritage for global audiences, it also sets up an immediate benchmark for what similar retrospectives could look like for other under-documented histories.