Dead man testifies in AI avatar
Chris Pelkey, a 37-year-old victim of a 2021 Arizona road rage shooting, utilised AI to resynthesise his voice and face for a victim impact statement in the case. The AI-produced video, created from Pelkey’s previous recordings and photographs, expressed forgiveness and was commended by the judge. “It is a shame we encountered each other that day in those circumstances,” the Pelkey avatar says in the video.
“In another life, we probably could have been friends.” Reports say some praised it as a forceful application of technology, others questioned the ethics of AI as a reflection of the deceased’s true desires. Pelkey’s sister Stacey Wales, however, said she scripted the AI-generated message after struggling to convey years of grief and pain in her own statement. She said she was not ready to forgive the accused, but felt her brother would have a more understanding outlook.
Overconfident, biased as humans
A recent study published in Manufacturing & Service Operations Management has found that AI chatbots such as ChatGPT share typical human decision-making biases. The researchers compared GPT-3.5 and GPT-4 on 18 well-known cognitive biases found in human psychology in a first to evaluate ChatGPT’s behaviour and discovered that “in nearly half of the test cases”, AI replicated irrational human decision-making.
“GPT-4 prefers certainty more than even humans do,” the authors say. The hot-hand fallacy and confirmation bias were especially pronounced. They found that though GPT-4 was superior on tasks that were logic-based, it struggled with subjective ones. “AI should be managed like a worker who has significant decision-making authority,” co-author Meena Andiappan has been quoted as saying in Live Science, emphasising the importance of control.
Books by AI
As per a report in The Guardian, Amazon is selling self-help books about ADHD written by AI that experts say are full of “dangerous nonsense.” Eight such books were found by Originality.ai to be 100% chatbot-written. King’s College London’s Michael Cook described it as “frustrating and depressing,” saying AI has the habit of combining pseudoscience with medical advice. Others slammed the absence of regulation, predicting a “race to the bottom.” One reader said he felt “upset” by the misleading content Amazon made money from.
First AI ad
Emami’s Dermicool has launched India’s first fully AI-generated FMCG ad, DermiCool Warriors, replacing typical summer visuals with a sci-fi fantasy where AI-created heroes battle prickly heat monsters. Developed in 30 days, the campaign blends nostalgia featuring its classic jingle aaya mausam thande thande DermiCool ka with AI visuals.
Emami says it aims to engage the new generation consumers while adopting modern storytelling formats and retaining the legacy they “are proud of”. It said the ad highlights AI’s cost-effective, agile storytelling potential in modern consumer marketing.
Wikipedia using GenAI?
As per a report in The Verge, Wikipedia is applying generative AI to aid, rather than substitute, its volunteer editors by automating processes such as background research, translation, and onboarding. The Wikimedia Foundation announced that it was integrating generative AI into its editing process as a means to help its volunteer and largely unpaid staff of moderators, editors, and patrollers reduce their workload and focus more on quality control. It has been quoted as saying that it stresses a “human-centric approach”, transparency, and adoption of open-source tools. Though AI is already used for vandalism detection and readability, it is said to be its first explicit support for editors. The development is meant to counteract increased bot traffic and alleviate the increased burden on human moderators and server resources.
Animal sounds in words?
Chinese tech behemoth Baidu is exploring the potential to use AI to interpret animal sounds into human language. As per a report in Sky News, a patent filed by Baidu by China National Intellectual Property Administration recently depicts a system that collects animal vocal, behavioural, and physiological information to identify emotional states, which could enhance cross-species communication. Though at the research phase, the concept has generated interest and doubt. The feasibility of applying it in real life is unclear, though technological progress is making such efforts more possible.??
Humanoids in manufacturing
China is leading the way in humanoid robotics, employing AI to transform manufacturing under economic pressures. In a warehouse in Shanghai, dozens of humanoid robots can be seen manoeuvred by their operators to carry out tasks like folding a T-shirt, making a sandwich and opening doors, over and over again. Startups such as AgiBot and MagicLab are combining hardware with AI to develop affordable, task-oriented robots. With huge government subsidies behind them, China leads the world in global production.
While deployment can displace workers, officials reportedly suggest solutions such as AI unemployment insurance and stress benefits in industries such as caring for the elderly.
Preventing credit card fraud
Mastercard is using AI to identify and prevent credit card fraud in evaluating patterns for almost 160 billion transactions per year. Its AI-driven platform, Decision Intelligence, provides real-time risk scores to transactions, eliminating false positives and identifying suspicious behaviour. It said it also utilises behavioural biometrics and other AI to fight fraud rings and customer deception, having better data security and trust.