The 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT 2025) will take place from June 23-26 in Athens, bringing together over 500 researchers, technologists, policymakers, and civil society leaders. This flagship interdisciplinary conference aims to address the social and technical dimensions of algorithmic decision-making amidst the rapid integration of AI technologies. The agenda features more than 250 sessions, including 217 peer-reviewed papers, workshops, and tutorials, focusing on ethical concerns in areas like healthcare, hiring, and criminal justice. Keynote speakers, such as Suresh Venkatasubramanian and Nathalie Smuha, will discuss the ethical implications of AI. The conference aims to foster a dialogue on governance, accountability, and the public good in AI systems. With support from major tech companies, FAccT 2025 underscores the urgent need for discussions on AI ethics alongside technical advancements. Media may attend virtually or in-person, with resources available online.
Source link
Global AI Ethics Leaders Convene in Athens to Forge a Responsible Future for AI
Maximizing Your Budget: Harnessing AI for Smart Savings
Using artificial intelligence (AI) can simplify budgeting and saving, making financial management less daunting. AI tools help users track spending, create personalized budgets, and find new saving opportunities based on individual habits and financial goals. General-purpose AI assistants like ChatGPT and dedicated apps like Cleo or Trim assist in generating shopping lists, comparing costs, and suggesting ways to manage expenses. Users can input their income, regular expenses, and financial goals to receive tailored budget recommendations. Automated budgeting apps handle tasks like expense tracking, while more advanced features in paid options may negotiate bills or identify unused subscriptions. Although AI provides valuable suggestions, it’s essential to verify the advice and refine prompts for better results. By incorporating AI into financial routines, individuals can make informed decisions, enhance their savings, and adapt their budgets over time, ultimately gaining greater control over their finances.
Source link
Navigating AI Hallucinations: Strategies for Addressing the Unsolvable Challenge
AI hallucinations are an inherent consequence of limited data and the probabilistic nature of large language models (LLMs). They arise when AI processes inputs that do not align with its training data, leading to outputs described as “hallucinations.” These models generate sequences based on patterns but lack true understanding. Consequently, they can produce errors that confuse users, leading to misconceptions about AI’s capabilities. Despite numerous attempts to minimize these hallucinations, complete resolution seems impossible due to their structural basis in generative AI architecture. This phenomenon has significant implications for data integrity and security; outputs require external validation. Moreover, while LLMs may excel in specific dense data areas, they falter outside these zones, making reliance on them perilous for critical applications. As models become more complex, hallucinations may also become more sophisticated, obscuring error detection even for experts, thus raising concerns about their impact on information accuracy and a potential feedback loop of misinformation in future AI training scenarios.
Source link
AI Song Recognition Feature Can Recognize Humming
Shazam is an essential app for quickly identifying catchy songs on iPhone, a feature that has been available on Android for years. However, in June 2025, Google’s Gemini AI introduced a new capability allowing users to identify songs simply by humming, whistling, or singing a melody. This feature enhances existing song identification technology in Android, previously provided by Google Assistant. Users can now click the mic icon and hum for 10-15 seconds to discover a song, which builds upon Google’s earlier “hum to search” launched in 2020. Additionally, Gemini allows users to upload video clips for analysis, making it easier to identify songs playing in the background. This dual-functionality promises to streamline music discovery while adding a layer of interactivity for users, whether they are at events or engaging with music online. The features are available across iPhone and Android, catering to a broader audience.
Source link
Complexity Scientists Call for a Revamped Blueprint for AI Development
Two Chinese researchers, Li Guo and Jinghai Li, argue that current AI systems, particularly neural networks, may be reaching a limit due to their logical disconnect from real-world complexity. In a new paper published in Engineering, they propose a paradigm shift towards aligning AI’s logical structures—datasets, models, software, and hardware—with the multilevel complexities of the systems they model. Their approach, based on principles from mesoscience, aims to enhance explainability and robustness in AI by integrating the principle of compromise-in-competition (CIC). This could lead to AI systems that operate more like well-calibrated telescopes rather than opaque black-box models. Adopting this framework could improve simulations in engineering, offer better transparency for AI safety, and reduce “hallucinations” in large language models. Overall, this vision emphasizes substance over mere scalability in future AI development, making a case for more coherent, real-world aligned architecture.
Google Unveils AI-Enabled Search Live Feature Powered by Gemini: Discover How It Works
Google has launched Search Live, a new voice-activated feature for its mobile app that allows users to engage in back-and-forth conversations with Google Search. Available for Android and iOS in the U.S., users can access this feature by tapping a “Live” icon and can ask questions verbally. Powered by the Gemini AI model, the system provides AI-generated audio responses along with relevant web links, facilitating natural follow-up queries without restarting the session. It also supports multitasking by operating in the background and includes a transcript button for easy reference. While leveraging Google’s reliable search infrastructure, concerns have arisen about the potential reduction of traffic to original sources, as users can access complete information directly. Future integrations will include visual search capabilities that allow users to interact with real-world visuals while conversing with the AI, expanding the functionality announced earlier at Google I/O.
Source link
Shareholders File Lawsuit Against Apple for Allegedly Inflated AI Advancements
Apple is facing a proposed securities fraud class action lawsuit from shareholders who claim the company misrepresented its capabilities in integrating advanced AI into its Siri voice assistant. This alleged downplaying negatively affected iPhone sales and stock prices, resulting in significant financial losses, potentially in the hundreds of billions. The lawsuit, filed in San Francisco federal court, targets CEO Tim Cook, CFO Kevan Parekh, and former CFO Luca Maestri. Shareholders, led by Eric Tucker, argue that at the June 2024 Worldwide Developers Conference, Apple implied AI would be crucial for the upcoming iPhone 16, yet the company lacked a viable AI-based Siri prototype. Concerns heightened after Apple announced delays in AI upgrades to 2026 and disappointing assessments during the recent conference on June 9. Since reaching a record high in December 2024, Apple shares have dropped nearly 25%, erasing around $900 billion in market value.
Source link
Midjourney Sparks Controversy with Disney Character Videos Amid Major Copyright Lawsuit
Midjourney has launched its new AI video tool, V1, capable of creating animated clips from user-uploaded images, even featuring copyrighted characters from Disney and Universal. Despite recent legal challenges from these studios claiming copyright violations—specifically citing outputs that involve characters like Wall-E and Yoda—testing revealed V1 can still generate animations of various protected characters, including Minions and C-3PO. While Midjourney implemented some restrictions to block certain prompts, these guardrails aren’t foolproof; users found ways around them by altering the prompts. This raises concerns as the lawsuit asserts that the video function exacerbates its ability to distribute infringing materials. Critics argue that Midjourney’s advancements in AI technology have not addressed the underlying issues of copyright infringement adequately, highlighting a growing tension between AI innovation and intellectual property rights.
Source link
MIT Student Develops AI-Enhanced Polymer Masks for Rapid Painting Restoration
MIT graduate student Alex Kachkine has developed an innovative AI-based technique to restore damaged paintings in hours, significantly speeding up traditional restoration processes that can take months. His method employs AI-generated polymer films that serve as transparent “masks” with thousands of color-matched regions, allowing conservators to apply them directly to original artworks without permanent alteration. This reversible process is documented digitally, providing future conservators with a clear understanding of the artwork’s restoration history. Kachkine’s inspiration stemmed from observing the backlog of damaged art during a cross-country drive to MIT, where he realized the immense cultural heritage hidden in storage. To illustrate his technique, he restored a 15th-century oil painting with 5,612 damaged areas in just 3.5 hours, offering a solution to the scarcity of skilled conservators needed for traditional restoration methods. His research highlights the potential of technology in art conservation, making previously overlooked artworks accessible once more.
Source link
Unlock Enhanced Gboard Features with Pixel AI for Instant Edits and More
Google’s Gboard is set to receive several new features, including the “SmartEdit” tool, which utilizes generative AI to assist users in editing text via voice commands. Code discovered by tipster AssembleDebug suggests that this feature will process edited text temporarily through Google. Additionally, a rumored AI feature called “Magic Cue” may enhance Gboard’s capabilities by allowing it to retrieve information, such as flight details, from connected apps without user intervention. Users may also find it easier to save and access custom emoji combinations in a favorites list. These updates aim to streamline user experience, particularly by minimizing app-switching through Gboard’s clipboard functionality. While these features are still in development, they indicate Google’s ongoing commitment to integrating AI into user-friendly tools. Expect significant updates to Gboard as these features evolve in tandem with advancements in Android.
Source link