AI Apocalypto Newsletter | Week 12
Ai News This Week! | The AI Advocate
Welcome to Week 12 of the AI Apocalypto Newsletter, exploring the latest in AI. This edition delves into the Generative AI Copyright Disclosure Act, AI's impact on mammogram analysis and antimicrobial resistance, and MIT's innovative chatbot testing method. We also discuss unethical AI use, including Microsoft's predictions for the 2024 elections.
Ethical Usage of AI
The proposed legislation aims to ensure fair use of copyrighted data in AI models.
Generative AI Copyright Disclosure Act
• Proposed by California congressman Adam Schiff.
• Requires AI model companies to disclose copyrighted training works.
• Penalties for non-compliance.
• Supported by media trade organizations.
• Balances innovation with creators' rights.
AI in Mammogram Analysis
• Enhances accuracy and speed.
• Highlights potential for improved cancer detection.
• Reduces false positives.
• Concerns about effectiveness across diverse populations.
• Questions on breast cancer survival rates.
AI Uses to Address Antimicrobial Resistance
• WHO warns of global health threats.
• MIT's Jameel Clinic uses AI to identify antibiotics.
• Semapimod found to target Gram-negative bacteria, opening new treatment avenues.
MIT's Machine Learning Technique for AI Chatbots
• Uses curiosity-driven exploration.
• Generates diverse prompts.
• Elicits toxic responses.
• Outperforms human testers.
• Enhances efficiency and scalability of AI model verification.
Unethical Usage of AI
AI tools used for cheating in education, prompting school bans.
Microsoft Threat Intelligence Predictions 2024 Elections
• Beijing's increased use of AI-generated disinformation.
• Focus on the US, South Korea, and India.
• Recent attempts in Taiwan's election.
• Low current impact, and future risk due to AI sophistication and investment.
Chief Justice of India's AI Perspective
• Highlights the transformative potential of AI in legal work.
• Stresses the need for ethical considerations to protect individual rights.
• Raises concerns about AI errors and bias.
• Warns against AI exacerbating legal inequalities.
• Notes potential to democratize legal access.
Hackers Use Facebook Ads for Fake AI Services
• Promotes fake AI services like OpenAI's Sora, DALL-E, and Midjourney.
• Tricks users into joining malware-laden communities.
• Steals sensitive data like credentials and credit card details.
• Users are advised to avoid untrusted downloads.
OpenAI's ChatGPT Disrupts Education
• Students cheating due to AI tools.
• Educators struggle to adapt.
• Some see AI as a collaboration opportunity.
• Concerns over accuracy and critical thinking lead to school bans.
Adecco Group Survey: AI Impact on Workforce
• 41% of global CEOs anticipate workforce reduction due to AI.
• AI's impact could lead to fewer job opportunities.
University of Chicago's Nightshade Tool
• Poisons AI image generators.
• Diffuses output when used without consent.
• Aims to protect copyright.
• Joins MIT's PhotoGuard in safeguarding artistic work.
AI Regulating Complexities
• Box CEO Aaron Levie advocates minimal regulation.
• Google's Kent Walker contrasts with Silicon Valley's approach.
• Generative AI Copyright Disclosure Act of 2024 addresses AI training copyright issues.
AI Advancements
A new multimodal AI model processes text, diagrams, and photos, achieving high benchmarks.
xAI Introduces Grok-1.5V: First Multimodal Model
• Processes text, documents, diagrams, charts, screenshots, and photographs.
• Competes with leading models in RealWorldQA benchmarks.
• Plans significant updates to Grok AI despite past controversies.
Marvell Technology Secures AI Chip Production Contracts
• Securing new contracts with major U.S. cloud computing companies.
• Custom chip units yield lower margins.
• Shares decline by 2.7% post-investor presentation.
US-Japan Collaborations in AI, Quantum Computing, Semiconductors
• $110 million AI research partnership.
• Advancements in quantum technology.
• Semiconductor cooperation.
• Development of human capital for the digital age.
Stability AI Launches 12B Parameter Base Model and Instruction-Tuned Variant
• Part of Stable LM 2 series.
• Models trained on two trillion tokens in seven languages.
• Aim for efficiency and strong performance.
• Aim to empower developers and ensure data control.
Adobe Launches Acrobat AI
• Gen AI-powered document summarization and search assistant.
• Available on desktop, web, and mobile.
• Priced at $4.99 per month until June 5.
• Offers insights, formatting, content sharing, and transcription features.