India’s Deepfake Dilemma

Can't read or see images? View this email in a browser

THE WEEKLY NEWSLETTER OF AIM. Sunday, Nov 26, 2023 | Was this email forwarded to you? Sign up here By Amit Naik A lot of you might argue that deepfakes spread misinformation, have the potential to ruin the reputation of individuals and organisations, and are a threat to democracy. In short, the technology does more harm than good, and is getting better each day. Agreed. Mainstream media, too, is going ahead with this narrative, spreading fear around deepfake and blaming social media and tech companies for the menace. The Indian government, which aligns with this view, has devised a so-called ‘clear, actionable plan’ to address the challenges posed by deepfake. While everyone's at it, let’s pause for a minute and talk about the ‘bright side of deepfake’ and why it should not be over-regulated.  So far, deepfakes have found actual use cases in movies and ads, mostly assisting in video editing and production. Some of the notable examples include Shah Rukh Khan’s My Ad - Cadbury campaign, Salman Khan’s double role - Pepsi ad, and Ayushman Khurrana’s latest ad with Wakefit, where the actor was showcased as a child with the help of deepfake. This ain’t new. Ten years ago, after the death of Hollywood actor Paul Walker, the team used deepfake to complete the Fast & the Furious 7 movie.  Besides movies and entertainment, deepfake also finds use cases in healthcare. In a 2018 study conducted by NVIDIA, Mayo Clinic, and MGH & BWH Center for Clinical Data Science, they used deepfake to improve AI diagnostic accuracy by supplementing rare abnormality images with synthetic ones. It can also help improve the performance of self-driving cars and autonomous vehicles. The list goes on.  Understanding Deepfake  Unlike photoshopped images and videos, which a large majority of people can now naturally identify flaws in, deepfakes are capable of creating a much more realistic, immersive, and impactful experience. Below is a glimpse of the Spanish AI-generated model, Aitana Lopez, created by the agency called The Clueless, which already earns nearly INR 3 lakh per month. “The tools for creating deepfakes will just get better and better in the future. They will get so good that it will be impossible to detect whether this was an original or fake,” said Stanford University professor and computer scientist Maneesh Agrawala, in an exclusive interview with AIM.  So, what now? From a user standpoint, when using tools to manipulate videos, two things need to be considered – the audience should be made aware of the manipulated video, and consent must be obtained from the actor in the manipulated videos. The same holds for images as well, and it becomes essential to have AI-generated labels or watermarkings.  How about developing a tool to detect deepfakes? “Such tools are not perfect. But we have algorithms and techniques to find those imperfections,” shared Agrawala.  He said that the emotions that forced the mouth to close often are not well reproduced by some tools. “That can work to some extent, but they are not foolproof,” he added.  To solve this problem, he, and his team at Stanford, have been working on developing a tool that considers lip movement around certain phonemes, where they were focusing on the visemes associated with words having the sound M (mama), B (baba), or P (papa), in which the mouth must completely close in pronouncing these phonemes.  Recently, IEEE researchers unveiled NPVForensics, which helps accurately detect deepfakes and outperforms state-of-the-art methods.  Regulating deepfake needs a delicate balance  While government supervision can prevent misuse and reduce ethical concerns, tightening the noose around deepfakes could hinder technological advancement and beneficial applications.  It’s time the Indian government proactively invested in developing models and algorithms that can detect deepfakes, instead of taking reactive measures, which are often easily bypassed or broken. AI Forum for India. Our Discord Community for AI Ecosystem, In collaboration with NVIDIA. Join Today >> BEST TECH STORIES OF THE WEEK >> xAI’s Grok Might End Up as A Joke It looks like Musk’s fear of legacy media and his attempt to enhance the engagement on X with the release of Grok next week is coming to life, as most publications are already integrating generative AI capabilities and features into their platforms, and who knows, xAI’s Grok might end up as a joke after all. The OpenAI Mahabharat  While the OpenAI fiasco came to an end (so, we believe) before Thanksgiving, people were quick to compare it to the Game of Thrones series with Sam Altman being Jon Snow. However, the real Altman was not a Snow who “knew nothing”, but was more like the courageous, skilled, intelligent archer Arjun from the epic Hindu mythology, Mahabharat. So, let’s find out who played whom in ‘The Mahabharat of OpenAI’. NVIDIA, the Real Winner in the OpenAI Fiasco While OpenAI's top leadership dealt with the boardroom drama, it was business as usual for the rest of the team. The servers kept buzzing with requests from millions of users and developers, and the team even shipped a new update for ChatGPT, making the voice feature available to all. In other words, more GPUs, and a whole lot of love from NVIDIA. Read the full story here. How Redis is Fueling the Generative AI Wave Redis has played an instrumental role, time and again, in helping ChatGPT keep the conversation going. In a recent interview with AIM, Tim Hall, the chief product officer at Redis, shares how it has been an integral part of many companies, not just OpenAI, in running their AI systems smoothly, and faster than ever before. Read the complete story here. AIM EVENTS >> DevPalooza  AIM is hosting an in-person developers’ meetup in Bengaluru on December 2, 2023. This is exclusively for the members of AI Forum for India, in partnership with Nvidia. Explore the journey into the LLM World.  MLDS 2024 The 6th Edition of Machine Learning Developers Summit 2024, is to be held next year (February 1-2, 2023) at the NIMHANS Convention Centre, Bengaluru. The conference will see over 1,500 in-person attendees, more than 50 speakers, and 400+ organisations expected to participate, reflecting the enormous scale and impact the event has on the ML community. Click here to reserve your spot now. AIM VIDEOS >> Welcome to "Beyond the Code," our exciting new YouTube series that takes you on a captivating deep dive into the multifaceted world of technology. In this series, we explore and discuss every aspect of the tech universe, unraveling the intricacies that make it so fascinating.  Here's the First Episode: AIM NEWS >> IBM opens its new IBM Client Innovation Centre (CIC) in Gandhinagar, India. Check it out here.  AWS India has introduced the ‘AI Ready’ initiative aimed at democratising AI education by offering free training to two million individuals by 2025. Genentech, a member of the Roche Group, and hardware giant NVIDIA have sealed a multi-year research collaboration for drug discovery.  Andrej Karpathy unveils a new course – ‘Intro to Large Language Models’-- based on his recent 30-minute talk on LLM at the AI Security Summit. Watch here.  OpenAI is secretly working on the next-generation AI system Q* (most likely built on Q-Learning and PPO), inching towards AGI.  Download our mobile app Access our stories, video content and events on the go. Stay Connected info@analyticsindiamag.com © 2023 Analytics India Magazine Pvt Ltd and/or its affiliates. All rights reserved. For more information, email info@analyticsindiamag.com

This email was sent by support@analyticsindiamag.com to alexvarboffin.abbb@blogger.com Not interested? Unsubscribe | Manage Preference | Update profile Analytics India Magazine | 280, 2nd floor, 5th Main, 15 A cross, Sector 6, HSR layout Bengaluru, Karnataka 560102