Twins

A unique group of senior Silicon Valley executives is using so-called AI “digital twins” to delegate part of their daily responsibilities. Meta already warned more than a month ago that they were preparing an AI version of Mark Zuckerberg so that employees could talk to their CEO, but little by little more cases have appeared. The concept is disturbing: an AI system analyzes how they speak, write and even how they think based on the history of emails, speeches and articles. From there, the digital clone He is capable of answering subordinates’ questions, writing proposals in his own style and even, in the most advanced cases, creating video avatars that give lectures in several languages ​​simultaneously. The example of Reid Hoffman. This executive, co-founder of LinkedIn and partner at Greylock Partners, is a clear reference in this trend. His clone, which he called Reid AI, was trained with 22 years of his own content (books, podcasts, articles) to provide it with all the necessary information. The concept of digital twin, by the way, had already been used in the technological field, but with another approach quite distinct. The one on the screen is not me, it is my digital twin. Since it began using it, this digital twin has given more than 75 presentations. In fact, in one of them this AI clone was presented to the public in French, Chinese and Hindi from a giant screen. The executive highlights that “I only speak one language, my AI speaks 74, ensuring that your AI digital twin saves you 50% of the time in the weeks in which you deploy it. For Hoffman, in a decade any company with more than 50 employees will assign trained virtual twins to its managers and middle managers. Not just for speeches. These types of bots are also getting fully involved in the area of ​​human resources and internal management tasks. Bala Sathyanarayanan, HR director at the multinational packaging company Greif, uses the so-called Balabot. This chatbot has interacted with more than 3,300 employees to resolve complex questions, such as motivating underperforming workers. Barriers. As in the case of Hoffman, this manager made use of public appearances and documents, but not those private and more sensitive ones: “He does not ingest my private email or confidential files,” he assured. in WSJ. The tool works so well, says Sathyanarayanan, that some company managers claim that several employees have redirected their careers thanks to the advice of their boss’s clone. But. These digital twins however have some problems. Kelly Monahan former director at Upworkhad to turn off his Digital Kelly clone live at a conference when she began “stuttering and repeating the same phrase on a loop” in front of 200 hoteliers. Hoffman also admitted that his AI sometimes goes completely deadpan after telling a joke, for example, breaking empathy with the audience. The lack of specific data is also another obstacle: Red AI was asked what his favorite ice cream was and he answered “vanilla” because he didn’t know the answer: Hoffman’s is chocolate. Rejection among employees. There is a clear enemy in this trend, and it is the rejection of employee templates. Analyst Josh Bersin attempted to integrate digital twins of employees so that AI could compose business emails by imitating their respective styles. The workers rebelled: “No one wanted to put their entire email history into the system.” Skepticism persists among them, but some claim have turned your digital clone into a daily assistant to prepare meetings or analyze market trends. What happens if you get fired. There is a dilemma more typical of an episode of ‘Black Mirror’ than our present. If an AI becomes brilliant at its job after absorbing all the accumulated experience and knowledge, can you take it with you on a pendrive if you change jobs? Lawyer Paul Jurcys explains that it is likely that in the near future companies will have to financially compensate departing employees so that they leave their digital twin and database behind. First I clone you, then I fire you? There is an uncomfortable question when talking about this topic: will companies create digital clones for the sole purpose of replace human workers and thus save their salaries? Gartner analysts already warn that doing so without transparent communication and without the employee’s explicit consent will cause notable social rejection. There are also doubts about what happens if that digital twin makes a serious mistake: whose responsibility is it, the human employee or the company that has integrated that virtual clone? You take your double, but not what you learned. Kelly Monahan lived this situation upon leaving the company. The employee retains the rights to his or her image, voice, and personal experience, but the company retains the proprietary data of the business that the AI ​​managed to capture during that stage. After leaving the company, Upwork deleted her digital twin, but she ended up retraining a “virtual double” independently with data from her next book to use in her next stage as an independent consultant. Image | Meta, Wikimedia Commons (Anthony Quintano) In Xataka | How to create a character in ChatGPT and Gemini to use it in all the images you make with artificial intelligence