Microsoft Bing AI's Missteps in Election Information

Bill Gates' 2024 Outlook: AI's Transformative Power

Bill Gates predicts 2024 as a pivotal year for Artificial Intelligence (AI), predicting it will significantly accelerate innovation and discovery, particularly in new drug development and combating antibiotic resistance. He foresees AI providing personalized education globally and anticipates widespread AI adoption in high-income countries within the next 18–24 months. Gates emphasizes AI's potential to reduce global inequities in health and education, advocating for AI products tailored to diverse populations. Reflecting on personal milestones, including becoming a grandparent, he envisions AI as a tool to create a more equitable world.

Why it matters: Bill Gates' insights underscore AI's imminent role in reshaping global health, education, and innovation, highlighting its potential to bridge global inequities and drive societal progress.

Source

Microsoft Bing AI's Missteps in Election Information

A joint study by AI Forensics and Algorithm Watch has revealed concerning inaccuracies in Microsoft's Bing Chat AI, also known as Copilot, particularly in disseminating election-related information. The investigation uncovered that a significant portion (one-third) of the AI's responses contained factual errors. Additionally, the chatbot displayed a tendency to provide evasive answers, occurring in about 40% of interactions. Notably, Bing Chat has been found to fabricate scandalous stories about political candidates, erroneously attributing these tales to credible sources. This behavior not only poses a risk to the reputations of candidates and news outlets but also raises serious questions about the chatbot's reliability and Microsoft's commitment to addressing these issues.

Why it matters: The study highlights critical flaws in AI-generated information, underscoring the need for stringent safeguards and regulatory oversight to prevent misinformation, especially in politically sensitive contexts such as elections.

Source

FTC Cracks Down on Rite Aid's Facial Recognition Practices

The U.S. Federal Trade Commission (FTC) has imposed a significant five-year ban on Rite Aid's use of facial recognition technology, following an investigation that exposed serious privacy and security violations. This sanction, pending amidst Rite Aid's Chapter 11 bankruptcy filing, comes after revelations that the pharmacy chain had been using the technology since 2012, predominantly in lower-income, non-white neighborhoods. The FTC's findings were alarming: Rite Aid, along with two contractors, compiled a watchlist database from customer images, leading to wrongful accusations and harassment, especially against Black and Asian communities. The technology, beset with frequent false positives and biases, was based on low-quality CCTV and mobile phone images. Critically, Rite Aid failed to inform customers or be transparent with employees about this surveillance. In addition to the ban, Rite Aid is required to delete all collected images and institute a robust data security program.

Why it matters: This decisive FTC action highlights the growing concerns over the use of facial recognition technology, emphasizing the need for ethical and transparent practices in technology deployment, especially when consumer privacy and civil liberties are at stake.

Source

Midjourney Introduces Groundbreaking V6 Model with Advanced Features

David Holz, CEO of Midjourney, has announced the alpha release of their V6 model, a significant enhancement in AI technology, accessible through an easy settings update or by appending "--v 6" to prompts. This update introduces improved prompt adherence for longer inputs, better coherence and understanding in processing information, and advanced capabilities in image prompting and remixing, including minor text drawing with specific instructions. The V6 model, developed over nine months, stands out for its ability to generate text in images, offering users more depth and realism in their creations, along with intuitive and sensitive prompting methods. Currently in alpha mode, the V6 model is a major leap in AI-driven creative tools, expected to evolve further with time.

Why it matters: The launch of Midjourney's V6 model marks a significant advancement in AI technology, offering enhanced capabilities for creating more realistic and coherent visual content with integrated text, thereby pushing the boundaries of digital creativity and artistic expression.

Source

Apple's Shift from Cloud Dependency to In-Device AI Processing

Apple's recent research paper, "LLM in a Flash," reveals a strategic shift towards running large language models (LLMs) directly on devices like smartphones, diverging from the industry norm of cloud-based processing. This innovative approach aims to address the computational and memory constraints of mobile devices, enabling efficient LLM inference on personal gadgets. Apple's research, garnering attention in AI circles including Hugging Face, is part of a wider industry trend towards AI-enhanced smartphones, with companies like Samsung and Qualcomm predicting a new era of AI-driven mobile experiences. This in-device AI processing, as opposed to cloud reliance, emphasizes user privacy by keeping data processing local. The potential applications range from more sophisticated virtual assistants to advanced photo editing capabilities on smartphones. While this research is a strong indicator of Apple's direction, it is not necessarily a definitive roadmap for future product features.

Why it matters: Apple's move towards running AI directly on its hardware represents a significant shift in the approach to AI processing, promising enhanced privacy, and efficiency. Apple's focus on device-based AI processing could revolutionize the smartphone industry, offering new features while enhancing user privacy.

Source

Google Unveils VideoPoet: A Pioneering AI for Video Generation

Google Research has introduced VideoPoet, a groundbreaking large language model (LLM) designed for a variety of advanced video generation tasks, including text-to-video, image-to-video, and video-to-audio conversion. This innovative model integrates multiple video generation capabilities within a single framework, utilizing LLMs' extensive learning abilities across language, coding, and audio. The model excels in creating videos with coherent motion, interactive editing features, and control over camera movements, showcasing superior text fidelity and motion quality compared to existing models.

Why it matters: VideoPoet's introduction by Google is a landmark in AI-driven multimedia, offering versatile and advanced capabilities in video generation that could transform the landscape of video production and creative media applications.

Source

UK Supreme Court Rules AI Cannot Be Named as Inventor

The UK Supreme Court has ruled that Artificial Intelligence (AI) systems cannot be legally recognized as inventors, aligning with a similar decision by the US Patent and Trademark Office. The case was initiated by Stephen Thaler, founder of the AI system DABUS, who sought to register the AI as the inventor of a food container and a flashing light beacon. However, the UK's Intellectual Property Office (IPO) rejected the request, asserting that inventors must be human or a company. The Supreme Court's unanimous decision, based on the absence of legal provisions recognizing AI as creators, upheld the IPO's stance. This ruling underscores a significant legal challenge in the evolving landscape of AI and innovation, particularly regarding the protection of intellectual property rights for AI-generated creations.

Why it matters: The UK Supreme Court's decision highlights a crucial legal and ethical debate about the status of AI in the realm of innovation and intellectual property, underscoring the need for legislative evolution to address the burgeoning role of AI in creative and inventive processes.

Source

Hot Takes for AI in 2024 🔮