In the year 2035, artificially-intelligent Agents trained on the personal data of individuals offer a compelling proxy for interacting with the people themselves. We feed these programs everything: old-fashioned text- and image-based data like email and social media posts, voice logs from smart speakers, gesture tracking from wearables, and even dreamscapes captured by experimental neural implants. Of course, these artificial Agents are only as good as their data sets, and tiny errors in transcription can have significant ripple effects.
Agent technology has transformed human interaction, and its reliability has become the crux of a landmark legal battle. Facing calls for increased regulation, neural implant maker Affect Link has mounted an aggressive defence in Australia’s High Court. The company’s hugely popular implants, which allow users to share physical sensations at a distance, have become a billion-dollar business despite their alleged health risks. The Commonwealth, citing growing concern over these risks, has asserted the implants fall within its constitutional power to regulate “postal, telegraphic, telephonic, and other like services”.
In a surprise move, a lawyer for Affect Link has called a digital Agent of Alfred Deakin, the second Prime Minister of Australia, as an expert witness in the case. The lawyer claims the agent, trained on notes from the constitutional convention debates and Deakin’s personal records held by the National Library of Australia, can represent the original intention of the founders of the Australian Federation. The Commonwealth argues there is no way to assure the accuracy of the agent or the underlying data. Whatever the outcome, the case stands to set an era-defining precedent.
The scenario above is adapted from Deakin Speaks, World Listens, one of several speculative vignettes included in Custodians and Midwives: The Library of Future, a special report written by the School of Cybernetics in partnership with the National Library of Australia. The technology depicted in the vignette—digital avatars of long-deceased people, empowered to speak on their behalf—is imaginary, but closer to reality than ever before. Recent advances in generative artificial intelligence have yielded machines that can converse in a persuasively human manner, down to impersonating specific individuals, making the Agent of Alfred Deakin a fairly modest extrapolation. With new advances arriving daily, AI could fulfil the promise underlying all archival technology, from the printed word to high definition video: direct communication with the dead.
The modern history of technology has long been haunted by this dream. The rapid industrialization of the late 19th century gave rise to a concurrent boom in spiritualism. Society was being radically reshaped by technologies like electric lighting and the railroad, which demolished centuries-old conceptions of time and distance. In a world where the telegraph could transmit a message across the Atlantic, communication beyond the veil seemed well within reach. People of the era flocked to mediums, who claimed to communicate with the spirit world through a coded language of knocks and raps. The similarity to the Morse alphabet of the telegraph was not accidental. Great scientists, authors, and politicians were among the devotees of spiritualism, including one Alfred Deakin, who served as President of the Victorian Spiritualists’ Union before becoming Prime Minister. These cultural memories, so alive in their day, are now the province of our libraries.
By the end of the 1920s, spiritualism fell out of fashion, but the dream of technological communion with the dead was soon channelled into another modern imaginary: science fiction. The genre most concerned with the future is filled with portals to the past. In the universe of Star Trek, an immersive environment called the holodeck allows people of the 23rd Century to mingle freely with figures from history. An episode of Star Trek: The Next Generation finds the android Data playing poker with Albert Einstein, Sir Isaac Newton, and Stephen Hawking, appearing as himself in a memorable cameo. In William Gibson’s cyberpunk novel Idoru, the protagonist Chia enjoys the enviable privilege of being tutored in music by a digital avatar of David Bowie (under a different name, for legal reasons). In Isaac Asimov’s The Immortal Bard, Shakespeare’s timeless understanding of humanity makes him uniquely suited for technological transference to 1950s America. Even so, he ends up failing a college class on the works of William Shakespeare.
‘Reanimating’ historical figures in contemporary times is no longer limited to authorial imagination. Digitized memories and digitized artefacts now form the technological basis for bringing history to life. Media theorist Friedrich Kittler wrote that the realm of the dead is as extensive as the storage and transmission capabilities of a given culture. We already have digital simulations of deceased actors appearing in movies and holograms of deceased musicians performing onstage. If Large Language Models like ChatGPT can ‘learn’ to speak like—and therefore presumably to speak for—national or international authorities, the way we preserve their utterances becomes increasingly consequential. Careful curation, long the remit of our libraries, can serve to guide us in the generation and navigation of these historical simulations.
Read more in the AI and libraries series:
The view from 1950’s Hollywood The age of artificially intelligent search