Microsoft’s recent filing of a US patent for the creation of what it describes as a ‘conversational chatbot of a specific person’ has given rise to media speculation over how the boffins at Redmond will build on the blueprint.
The patent (No. 10,853,717 B2, dated December 01 2020) conceptualizes technology that would access ‘social data’ about that specific person, such as images, voice data, social media posts, emails and other authored documentation, that would instil personality and inform conversations between the bot and a live human.
‘In some aspects,’ the patent abstract continues, ‘a 2D or 3D model of a specific person may be generated using images, depth information, and/or video associated with the specific person’.
How this could all come together is outlined at length in the remainder of the patent’s 21 pages. An AI component is not specified in detail, but it’s hard to imagine such a concept could function without some AI to breathe life into it, so to speak.
Meanwhile, the patent and the possibilities its successful realization would deliver, have prompted some searching responses from online media commentators.
Barry Collins at Forbes reckons Microsoft’s patent “raises the intriguing possibility of digitally reincarnating people as a chatbot”. It’s an idea that also “raises all manner of privacy implications that aren’t discussed in the patent,” Collins continued. “Will people be given the right to opt out of such a system, for example? Would the relatives of the dead be able to prevent others from turning their deceased loved ones into chatbots?”
Collins added: “Such questions are, of course, moot until Microsoft – or someone else – delivers a working prototype. But it might not be the case for much longer that your personality dies with you.”
Andrew Paul at Input reckons the patent’s scope “sounds like Microsoft hopes to better improve its customer service chatbots, as well as future AI assistance along the lines of Alexa, Siri or its own Cortana”. However, Paul found the implications for this kind of hyper-specific, digitalized personality mimicry as varied as they are unsettling: “Aside from the very obvious creepiness of a deceased loved one reading you Waze directions to the nearest Chipotle from beyond the grave, the idea of conversational AI based on an amalgamation of your tweets, Facebook posts, and text messages, coupled with voice mimicry, could usher in a whole new era of cyber-fraud and ‘identity’ theft.”
Tyler Lee, writing at Ubergizmo, agrees: “By using AI and feeding it information, it could lead to next-generation levels of identity theft. We admit that it might be a bit nostalgic to be able to ‘talk’ to a dead loved one online where the chatbot can mimic that person realistically, but it still feels kind of weird – and a little bit wrong.”
Become an AI Forum Member
Did you enjoy this content? Join the AI Forum community today and subscribe to our newsletter to receive the latest AI news and relevant research. Don’t forget to leave a comment with your thoughts below.
