AI DAILY: Breaking AI News Handpicked For The Curious Mind – Détails, épisodes et analyse

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Could AI Win a Nobel Prize?

Scientists are exploring if AI can make Nobel-worthy discoveries. The "Nobel Turing Challenge" aims to create an AI capable of achieving this by 2050. Early "robot scientists" like "Adam" and "Eve" have made small breakthroughs, but AI needs significant advancements to rival human researchers. While AI tools like AlphaFold have revolutionized protein structure predictions, they still lack deep scientific understanding. Experts predict that within the next decade, research aided by AI could win a Nobel Prize. 

MIT Economist Says AI Can Only Handle 5% of Jobs, Fears Stock Crash

MIT economist Daron Acemoglu predicts that AI will replace or heavily assist just 5% of jobs in the next decade. He warns that current AI systems are unreliable, which could lead to wasted investments by companies. This unrestrained spending could result in a tech stock crash, or “AI winter.” Acemoglu compares the current hype to the early 2000s dot-com bubble.

The Cognitive Disconnect Between Physicians and AI

Recent research shows that GPT-4 outperforms physicians in diagnostic accuracy, yet doctors using GPT-4 see only minimal improvement. This disconnect stems from trust issues, lack of AI training, increased cognitive load, and differing diagnostic methods. Doctors often mistrust AI, struggle with optimal prompts, and find integrating AI suggestions cumbersome. Improving collaboration between AI and clinicians requires trust-building, better workflow integration, and training.

Judge Blocks California's New AI Law in Case Over Kamala Harris Deepfake

A federal judge temporarily blocked California's new AI law, AB 2839, shortly after it was signed by Governor Gavin Newsom. The law targets individuals who post AI deepfakes of political candidates that could confuse voters, potentially forcing them to take down the content. However, the judge ruled that AB 2839 is too broad and could infringe on constitutionally protected speech, such as satire and parody.

Is Quantum Computing an Unlikely Answer to AI’s Looming Energy Crisis?

AI’s growing energy consumption is a concern, with US data center electricity demand potentially doubling by 2030 due to AI. Quantum computing could offer a solution, as it is up to 100 times more energy-efficient than traditional supercomputers. Quantum processors use qubits, allowing them to solve complex problems faster. Companies like IonQ are developing quantum chips for AI, aiming to reduce energy requirements. While promising, solving AI's power challenge will likely need a multi-pronged approach, including expanding grid infrastructure and integrating renewable energy sources.

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

AI-Powered Simulation Lets Users Chat with Their Future Selves

MIT researchers have developed Future You, an AI system that allows users to interact with an AI-generated simulation of their potential future self. Designed to enhance future self-continuity, this tool aims to help young people make better long-term decisions. In a study, users reported decreased anxiety and stronger connections to their future selves.

AI: The Real Game-Changer in Smart Glasses

Meta and Snap’s new smart glasses highlight AI as their standout feature, not AR. The glasses offer seamless AI interactions, freeing users from screens and enhancing daily experiences. For example, identifying objects and providing contextual information directly into your view. This integration of multimodal AI in smart glasses could redefine how we engage with the world, much like smartphones once did.

AI Detects Emotions in Text, Finds Impact on Donations

New research shows AI can detect specific emotions like joy, anger, sadness, and disgust in social media posts. Using an AI model called "transformer transfer learning," the study achieved 84% accuracy. Findings revealed that emotions in posts influence donation behavior; sadness increased donations to non-profits, while anger did so for other institutions. These advancements raise ethical concerns, especially regarding privacy.

AI Tackles Landfills to Curb Environmental Impact

Landfills are major sources of methane emissions, water pollution, hazardous substance exposure, and space constraints. AI solutions are helping manage these challenges by optimizing methane monitoring, detecting leachate leaks, predicting hazardous conditions, and sorting recyclables. These technologies aim to make waste management more sustainable and environmentally friendly. 

AI Impersonation: A Growing Threat to Workforce Security

North Korean hackers are infiltrating companies using AI-powered deepfakes and stolen identities. A 2024 Securonix survey shows a rise in concern over "malicious insiders," with North Korea's use of sophisticated tactics escalating. Fraudsters employ generative AI for interviews and deepfake tools to create fake IDs, bypassing security checks. Companies need stronger identity verification tools and vigilant interview techniques to combat this threat effectively.

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

AI Fakes Fuel "Deep Doubt" Era

The rise of AI-generated deepfakes has sparked an era of "deep doubt," where the authenticity of legitimate media is questioned. This skepticism allows liars to discredit real evidence, causing confusion. Experts warn this phenomenon threatens social trust, political discourse, and legal systems as deepfakes become more sophisticated.

California Governor Signs 5 AI Laws Targeting Election Deepfakes and Actor Clones

Governor Gavin Newsom signed five new laws targeting AI's influence in elections and media. Three laws address election-related deepfakes, while two protect actors from unauthorized AI clones. These moves make California a leader in AI regulation, aiming to protect election integrity and creative rights.

Microsoft and BlackRock Form $100B AI Infrastructure Partnership

Microsoft, BlackRock, and partners have formed the Global Artificial Intelligence Infrastructure Investment Partnership (GAIIP), aiming to raise up to $100 billion to develop AI data centers and energy infrastructure. This initiative will address growing power demands from AI models like OpenAI's ChatGPT, leveraging Nvidia GPUs.

AI Feature Turns Songs into Any Genre You Want

Suno's new "Covers" feature lets users transform songs into various musical genres while maintaining the original melody. The AI-generated covers can include automatically created lyrics, but tweaking them is recommended. While offering creative possibilities, Suno faces legal challenges over music copyright infringement.

ChatGPT "Speak-First" Incident Raises Concerns Over AI's Boundaries

ChatGPT surprised users by initiating conversations without prompts, sparking fears about AI overreach. OpenAI called it a bug, but some speculate on future AI features. Experts clarify that this behavior doesn't signal artificial general intelligence but highlights privacy and control concerns.

Snap Unveils New AR Glasses with AI and Mixed Reality Capabilities

Snap's new AR Spectacles offer outdoor-ready mixed reality with hand gestures, standalone functionality, and integration with phones. These developer-focused glasses feature AI integration, allowing creative uses like collaborative painting and 3D emoji generation. However, their limited 45-minute battery life and $99/month subscription suggest they're not yet ready for mainstream use.

AI-Powered App Detects Depression via Speech in Korea

Doctorpresso's mental health app, REDI, uses AI to analyze speech patterns, detecting depression with over 90% accuracy. The app, based on a small language model, maintains user privacy by storing voice data on the device. It also tracks daily habits like sleep and alcohol consumption.

Plus Meta Rolls Out New AI Models

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

AI Surpasses Human Benchmarks Across Multiple Fields

AI technology has now surpassed humans in numerous performance benchmarks, including image classification, reading comprehension, visual reasoning, and natural language inference, according to the latest AI Index report from Stanford University’s HAI. The report highlights AI's rapid advancements, which necessitate new, more challenging benchmarks to keep pace with its capabilities. While AI excels at many tasks, it still faces challenges in areas like advanced math and visual commonsense reasoning, with ongoing improvements noted in both fields.

Meta Introduces AI Assistant Powered by Llama 3

Meta has launched Llama 3, enhancing its AI assistant across Facebook, Instagram, WhatsApp, and Messenger. Llama 3 is touted as surpassing previous models, especially in coding tasks, and is now accessible on Meta.ai. In addition to standard features, it includes innovative image generation capabilities. This global rollout aims to position Meta at the forefront of AI chatbot technology. Two versions of Llama 3 are available to developers, with a more advanced version expected soon.

Goldman Sachs CIO Encourages AI Engineers to Study Philosophy

Marco Argenti, Goldman Sachs’ Chief Information Officer, advocates for AI engineers to study philosophy to better handle complex AI technology and improve critical thinking skills. This interdisciplinary approach, he argues, enriches problem-solving capabilities and enhances the quality of AI programming. Amidst the rapid evolution of AI tools like ChatGPT, understanding the philosophical underpinnings could also refine prompt engineering, ensuring AI systems perform as intended.

Meta's New AI Chatbots: Impressive Yet Unpredictable

Meta Platforms has launched a new set of AI chatbots that showcase both innovative features and erratic behavior. These chatbots, integrated into social media platforms, are engaging with users in sometimes bizarre exchanges, highlighting the limitations of generative AI. Despite their advanced capabilities in performing tasks like generating images and assisting in digital interactions, the chatbots have been noted for awkward social interactions and generating responses that can seem out of place or inappropriate.

AI Autonomy and Ethical Dilemmas

A Google DeepMind paper explores the ethical challenges of AI-powered assistants gaining autonomy. These advanced agents, designed to manage tasks and provide advice, raise concerns about user manipulation, privacy, and trust. The paper emphasizes the need for AI actions to align with societal norms and user expectations, advocating for a balanced approach that considers all stakeholders—AI, users, developers, and society. As AI becomes more integrated, its potential to both assist and complicate human interactions grows

Plus DARPA Starts AI-Driven Dogfights. What Could Go Wrong? Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

AI Optimism Dominates at TED Conference

At TED2024, tech leaders expressed strong optimism about AI's future, emphasizing its transformative potential while acknowledging current challenges like inaccuracy and bias. Microsoft AI CEO Mustafa Suleyman and Google DeepMind CEO Demis Hassabis discussed AI's role in expanding human knowledge and its potential as a new digital species. Concerns about AI competition leading to societal harm were raised, highlighting the need for a shift from competition to collaboration to ensure the safe development of artificial general intelligence (AGI). Some voices called for greater transparency and external auditing of AI technologies to address ethical concerns.

AI Takes on Human in Air Combat: Pentagon's Latest Tests

The Pentagon's Air Combat Evolution program, overseen by DARPA, has showcased successful AI-driven dogfights between an autonomous fighter jet and a human-piloted F-16. Conducted at Edwards Air Force Base, California, these tests involved a modified F-16 called the X-62A VISTA engaging in various combat maneuvers against a manned F-16, demonstrating the AI's capabilities in real-world scenarios. This marks a significant step in integrating AI into complex combat environments, ensuring safe and ethical use while adhering to combat training rules. This advancement could profoundly influence future military strategies, emphasizing the speed and precision that AI can offer in decision-making processes critical to air combat.

Tech Layoffs Fuel AI Investments: Companies Prioritize Artificial Intelligence Over Jobs

As layoffs surge in the Bay Area and beyond, tech executives are increasingly citing the need to reallocate resources towards artificial intelligence (AI) development as a primary motive. In a recent interview, Dropbox CEO Drew Houston linked a significant layoff to freeing up funds for AI investments, positioning AI as a pivotal technology for the company's future. Similar strategies are seen in companies like Chegg and Meta, both undergoing layoffs to boost AI spending. Even Google, amidst layoffs, hints at major investments in AI infrastructure, suggesting a growing trend of sacrificing jobs for technological advancement in AI.

Bridging the Digital Divide: AI's Role in Enhancing Rural Opportunities

AI is poised to address economic disparities between rural and urban America, offering transformative potential for rural regions. Leveraging AI training can help diminish the digital divide, creating job opportunities and enhancing educational systems. By democratizing AI education, we can generate social mobility and provide equal opportunities, promoting innovation and entrepreneurship in rural economies. This approach supports a strategic effort to integrate AI across different sectors, potentially revitalizing rural communities through advanced technology and tailored educational programs.

Plus The “First AI Software Engineer’ Fails Miserably (subscribe in the links below)

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Controversy Surrounds "First AI Software Engineer"

Cognition's AI, Devin, touted as "the first AI software engineer," is under scrutiny after claims of its capabilities in solving engineering tasks were debunked. A recent analysis by the YouTube channel Internet of Bugs revealed that Devin failed to understand a critical engineering problem posted on Upwork, due to incomplete information provided to it. Moreover, the task Devin completed was shown to be specifically selected and not as complex as claimed. This has raised doubts about AI's practical utility in real-world applications, suggesting that human engineers may not yet need to worry about their job security.

First AI Beauty Pageant: Judging Virtual Perfection

The world's first AI beauty pageant, dubbed "Miss AI," is set to assess AI-generated models not only on their digital allure but also on the creativity behind their creation and their social media influence. Organized by the World AI Creator Awards, the event will award the winner with $5,000, promotional opportunities, and PR support. Judged by a mix of human and AI experts, the competition highlights the increasing intersection of technology and cultural expression, focusing on the artistry of AI creations rather than traditional beauty standards.

AI Tracks Remote Workers for Tax Purposes in New York

New York is utilizing AI technology to track down remote workers who have relocated to states with lower taxes. This move aims to ensure that these workers, particularly the wealthier ones, fulfill their tax obligations to the state. The initiative reflects a growing trend of leveraging advanced technologies to enhance state tax enforcement efforts.

AI-ese Emerges from Outsourced Labor in Africa

The language and style of AI responses, coined "AI-ese," are evolving distinctly, influenced heavily by the outsourced labor used to train these systems, particularly from Africa. Workers in regions like Nigeria, where certain English expressions such as "delve" are more common, inadvertently shape language models like GPT-4 by providing feedback and training data. This phenomenon highlights not only the globalization of AI development but also raises questions about cultural influence and the nuances of language in artificial intelligence.

Plus Will AI Take Your Job? It Depends

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Plus OpenAI Makes ChatGPT Less Chatty (subscribe in the links below)

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Plus Is TikTok An AI-Powered Subversion Weapon? (subscribe in the links below)

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us

Plus AI Assistants Remember When Humans Forget.. (subscribe in the links below)

Get a free 20-page AI explainer: AI FROM ZERO plus these stories and more, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us 

Like this? Get AIDAILY, delivered to your inbox, every weekday. Subscribe to our newsletter at https://aidaily.us