Plus, can AI reduce toxicity on social media?
AI researchers and experts are increasingly concerned about the lack of transparency and openness surrounding advanced AI systems like GPT-4. A study conducted by Stanford University indicates that major AI models, including those developed by OpenAI, Google, and Amazon, need to provide sufficient information about their data sources, training methods, and energy consumption. The lack of transparency reduces reproducibility, hinders scientific advancements, and raises concerns about safety and accountability. At the same time, some companies are making efforts to develop more transparent AI models, the industry as a whole needs to address these issues to ensure the responsible development and deployment of AI technology.
The chiefs of the FBI and Britain’s MI5 have raised concerns about the potential of AI in the hands of terrorists, emphasizing the need for international partnerships and cooperation between the public and private sectors. The Five Eyes Alliance conference, attended by the United States, United Kingdom, Canada, Australia, and New Zealand, focused on addressing emerging intelligence threats, particularly in relation to AI generative platforms. The FBI and MI5 directors highlighted the use of AI to amplify terrorist propaganda, hide alarming searches, and exploit AI-built infrastructure security. Experts advocate for partnerships between governments and private companies to combat these issues and harness AI for positive outcomes.
According to analyst Jeff Pu, Apple plans to introduce generative AI technology on iPhones and iPads as early as late 2024. The company is expected to build AI servers and offer a combination of cloud-based AI and edge AI, prioritizing customer privacy. Apple’s efforts in generative AI have been considered behind its competitors, but the implementation of large language models into Siri could be part of the iOS 18 update. Generative AI has gained popularity, and Apple has also confirmed its interest in the technology.
IBM has developed a brain-inspired computer chip called NorthPole, which is capable of running AI-powered image recognition algorithms at a speed 22 times faster than other commercial chips. This innovation also boasts 25 times the energy efficiency of comparable chips. Inspired by the human brain, NorthPole combines computation and memory into a digital architecture compatible with current silicon chip technology. Although specialized for running AI processes, the chip’s design limitations may be overcome by using multiple NorthPole chips to support larger language models. While the prototype may not be available commercially immediately, this breakthrough has significant implications for the efficient implementation of AI in self-driving vehicles and aircraft.
Google’s latest Pixel phones come with new generative AI photo-editing tools that make it easier to perfect our photos and subtly rewrite small details from our lives. While AI photo tools like Magic Editor may enhance our images and make them appear more vivid, it raises questions about the authenticity and accuracy of our memories. As AI continues to develop, it may alter the way we understand and remember the past, potentially creating a camera roll that is more like virtual reality than a true reflection of our lives.
A study conducted by researchers at the University of Amsterdam reveals the potential of using AI-powered chatbots to reduce toxicity in online conversations on social media platforms. The experiment involved 500 chatbots with different personas interacting with each other. Preliminary results show that the chatbots’ ability to bridge the gap between people with diverse political views could lead to better conversations and decreased polarization. This innovative algorithm offers the possibility of creating a less toxic social media environment.
A groundbreaking collaboration between Iraqi and Australian researchers has successfully combined the ancient practice of Chinese tongue diagnosis with artificial intelligence and machine learning technologies. By analyzing tongue images captured using a USB web camera, a computer-assisted diagnostic system accurately identified diseases such as diabetes and renal failure in 94% of cases. These innovative methods offer a cost-effective and remote health monitoring alternative, especially during pandemic-restricted scenarios. This study highlights the remarkable potential of AI-driven tongue analysis in disease diagnosis.