📰 AI News, May 23rd 2024: AI Innovations and Ethical Dilemmas, Navigating the Future of Technology
Filipe Bento
Applied AI (ML, CV, GenAI), Cloud expert. PhD, MSc, Eng, Manager, Developer, Researcher | Head of Digital Resources & User Support, Libraries @UAveiro | Data ETL, Web tech, AI Agents Orchestration & Automation.
In this edition, I cover 42 entries showcasing the latest advancements and challenges in AI, from Microsoft’s AI PCs at Build 2024 to ethical concerns over voice cloning and AI’s impact on management.
ToC
Innovation and Development in AI
Microsoft Build 2024 : The Year of AI PCs
ChatGPT for Mac : Advancing Human-Computer Interaction
Open AI Goal and Strategy : Technology Sharing
Snapchat and Artificial Intelligence : Revitalization through Personalized Tools
Canva and Innovation with AI : Transforming the Creative Process
AI Innovations at Google I/O 2024 : Improvements in Generative Models and Tools
Copilot to Evolve : Reviews and Comparison with Other AI Bots
Apple and the Age of Artificial Intelligence : Challenges and Strategy
Artificial Intelligence and the Future of Innovation : Impact on Science and Technology
ChatGPT Mobile Application Revenue : Historical Peak with Launch of GPT-4o
Introducing Copilot+ PCs : New Processors and AI Models
Nadella's AI Era at Microsoft : Regulatory and Innovation Challenges
Fundamental Models of Multimodal Early Fusion : Understanding and Generating Images and Text
Artificial Intelligence and the Future of Web Search : Integration and Implications
Artificial Intelligence in Management : Transformation of Companies
'Small' Language Models : Pursuit of Big Profits
The AI Revolution Is Coming : Accelerating Innovation and Transformative Impact
Smart Glasses Revolution : Google's Project Astra and Meta's Multimodal Capabilities
Ethics and Society
Imperfect Clone of Sky's Voice : Discussion on the Value of Unique Voices
Privacy at Risk with Slack : Using Data for Model Training
Intrusion of Chatbots in Online Communities : Ethical and Social Concerns
AI Threats to the Entertainment Industry : Lawsuit for Misuse of Voices
Voice Selection for ChatGPT : Process Involving Casting Professionals
AI voice generates controversy : Similarity to Scarlett Johansson
OpenAI Superalignment Team Disbanded : Ethical Challenges and AI Security
Google DeepMind Border Security Framework : Protocols to Mitigate Risks
Regulation of AI as Crony Capitalism : Proposed Regulation
Data Property Rights : Solution to Runaway AI
Technology and Practical Applications
Work Intensive Culture : Resurgence in AI Startups
Artificial Intelligence in Accounting : Automation of Tasks and Financial Data
Retool Integration with Amazon Bedrock : Building AI Applications
Detailed Implementation of Llama3 : Advanced Programming Techniques
Gemini 1.5 : Generational Leap in AI Models
What's new in ChatGPT : Interactive Tables and Integration with Drives
Inflection AI : Emotional Intelligence for Enterprise Chatbots
Investigation and Security
Big Week for AI : GPT-4o and Gemini Find Their Voice
Reflections on Responsible Scaling Policy : AI Model Risk Management
Interim Report on the Security of Advanced AI : Risk Management and Diversity of Opinion
Disintegration of the OpenAI Superalignment Team : Fundamental Divergences
Gemini 1.5 : Near Perfect Recovery on Long Context Tasks
Market and Economy
The Retail Customer Service Revolution with Generative AI : Robust Growth Despite Economic Challenges
Artificial Intelligence and Identity Manipulation : Digital Identity Transformation
let's start
💻
Microsoft Build 2024: The Year of AI PCs
Microsoft will reveal how Windows will enable PCs with AI at the Build conference. PCs with new NPUs chips will enable AI tasks such as language translation in real time.
🔗Microsoft Build 2024 preview: The year of AI PCs (cnbc.com)
💬
Chatting with AI - The ChatGPT App for Mac
The ChatGPT application for Mac represents an advancement in human-computer interaction, allowing natural conversations and organization of unstructured thoughts. It faces challenges like verbosity and interruptions in voice capture, but it anticipates a future where tasks like composing emails will be done by voice.
🔗Chatting With Her - The ChatGPT App on Mac by @ttunguz (tomtunguz.com)
🗣️
Imperfect Clone of Sky's Voice
Benjamin De Kraker, a Twitter user, revealed that he created an imperfect clone of Sky's voice before it was removed. The version he shared is of lower quality than the original. Other users commented on the surprising similarity to the original voice.
The discussion evolved into the potential value of unique voices, with one user suggesting the situation could evolve into a billion-dollar bubble. Another user noted that many women sound like Scarlett Johansson, questioning the uniqueness of vocal impressions.
🔗Benjamin De Kraker 🏴☠️ on X: "Here's the official statement released by Scarlett Johansson, detailing OpenAI's alleged illegal usage of her voice... ...read by the Sky AI voice, because irony.https://t.co/cJDlnA0hTP" / X
🤖
Goal and the Open AI Strategy
Meta is committed to sharing its advanced AI technology for free, an unprecedented move in the sector. The company invests in high-performance chips and cutting-edge scientists, but unlike its competitors, it offers results at no cost.
🔗Facebook Parent’s Plan to Win AI Race: Give Its Tech Away Free - WSJ (archive.is)
💤
Work-intensive culture returns with AI startups
Work-intensive culture is resurfacing in Silicon Valley, with AI startups embracing nap pods and hacker houses. Exa Labs, an AI research startup, is one example, offering nap capsules to improve productivity.
🔗With AI startups booming, nap pods and Silicon Valley hustle culture are back | TechCrunch
🤖
Snapchat and Artificial Intelligence
Snap Inc. bets on artificial intelligence to revitalize Snapchat, focusing on personalized and adaptive tools. Evan Spiegel, the company's CEO, recognizes the need for innovation in the area of machine learning to maintain relevance in the market.
🔗Snap CEO Evan Spiegel Focused on AI, Machine Learning Push - Bloomberg
🔒
Privacy at Risk with Slack
Slack uses user messaging data to train machine learning models without explicit permission. The deletion process is complicated, requiring organization administrators to contact Slack.
🔗Slack has been using data from your chats to train its machine learning models (engadget.com)
🤖
Artificial Intelligence in Accounting
Accounting is facing a talent drain and AI is emerging as a solution, automating tasks and providing financial data in real time. The startup Puzzle uses AI to help startups manage finances and avoid failures due to inappropriate financial practices.
🔗AI Is Taking Over Accounting Jobs As People Leave The Profession (forbes.com)
🤖
Intrusion of Chatbots in Online Communities
AI chatbots are invading online spaces, replacing human interactions in virtual communities. Ethical and social concerns stand out, with an emphasis on the importance of human connection.
🔗AI chatbots are intruding into online communities where people are trying to connect with other humans (theconversation.com)
🧙
Canva and Innovation with AI
Canva, a visual communication platform, integrates AI to transform the creative process. Magic Studio, with tools such as Magic Write and Magic Design, facilitates the creation of visual content, reaching more than 5 billion uses.
🔗Canva's AI-powered Magic Studio Used 5 Billion Times and Counting | OpenAI
🛍️
The Retail Customer Service Revolution with Generative AI
Generative AI is transforming customer service in retail, promising robust growth despite economic challenges. Uses advanced language models to personalize the shopping experience and optimize the value chain.
🔗Game On, Retailers: Elevate Your Customer Experience with GenAI (cohere.com)
🤖
AI Innovations at Google I/O 2024
Significant improvements to AI models and generative tools revealed. Highlights include Gemini 1.5 with a context window of 2 million tokens and Trillium TPU, 4.7x more powerful than its predecessor.
🔗Google I/O 2024: 100 things Google announced (blog.google)
🚀
Copilot Evolving, But with Criticism
OpenAI's Mustafa Suleyman announced that Copilot will evolve to see, hear, speak and help in real time. The goal is for the AI assistant to start living life alongside the user, whether it's playing Minecraft or helping navigate life's toughest challenges.
However, one Twitter user, Arlo Gilbert, criticized Copilot for not being good enough at answering questions. He compared Copilot with other AI bots, such as Gemini, 4o and Claude, stating that they are much better at answering questions.
🔗Mustafa Suleyman on X: "We are taking Copilot to the next level. 🚀 Copilot will see, hear, speak and help in real time. Watch this demo to see what I mean. Soon your AI companion will start to live life alongside you, whether playing Minecraft or helping you navigate life’s most difficult challenges.https://t.co/ij6p3CgeYG" / X
🍏
Apple and the Age of Artificial Intelligence
Apple faces challenges in the generative AI race and needs to evolve its strategy. The need to differentiate the iPad from the Mac and the commitment to the success of the Vision Pro in Asia stand out.
🔗What Is Apple Doing in AI? Summaries, Cloud and On-Device LLMs, OpenAI Deal - Bloomberg
🤖
Artificial Intelligence and Identity Manipulation
AI transformed a Ukrainian YouTuber into a Russian figure, changing his digital identity. AI's ability to create convincing and potentially misleading representations stands out.
🔗How AI turned a Ukrainian YouTuber into a Russian (bbc.com)
🎙️
AI Threats to the Entertainment Industry
Actor couple sues company for misuse of AI voices, alleging violation of privacy and trademark laws. Voice cloning technology raises questions about copyright and impact on the job market.
🔗Voice Actors Sue Company Whose AI Sounds Like Them - The New York Times (nytimes.com)
🎙️
Voice Selection for ChatGPT
The selection of ChatGPT voices involved casting and directing professionals, with more than 400 applications evaluated before choosing the final 5 voices. The process lasted five months and included professional voice actors, talent agencies, casting directors and industry consultants. Actors are compensated above market rates and privacy is maintained without disclosing names.
🔗How the voices for ChatGPT were chosen | OpenAI
🎙️
Big Week for AI: GPT-4o and Gemini Find Their Voice
It was an important week in the world of AI, with OpenAI and Google releasing significant updates. Consumer partners Bryan Kim and Justine Moore discuss fellow multimodals who have found their voice.
It also discusses why not all audio is the same and why various nuances, such as speed and personality, really matter.
🔗A Big Week in AI: GPT-4o & Gemini Find Their Voice | a16z Podcast (simplecast.com)
🔒
Reflections on the Responsible Scaling Policy
Explores the implementation of a policy to manage risks from AI models, highlighting the importance of a structured framework and interdisciplinary collaboration. It reveals challenges in defining “Red Line Capabilities” and assessing emerging risks.
🔗Reflections on our Responsible Scaling Policy \ Anthropic
🧠
Interim Report on the Security of Advanced AI
Explores the evolution and risks of general AI, highlighting the need for risk management and the diversity of opinions among experts. Evaluates technical methods to mitigate risks, emphasizing that there are no absolute guarantees against potential damage.
🔗international_scientific_report_on_the_safety_of_advanced_ai_interim_report.pdf (publishing.service.gov.uk)
🚀
Retool integration with Amazon Bedrock
Retool announces integration with Amazon Bedrock, allowing you to build AI applications safely and quickly. The collaboration combines Amazon's AI model management with Retool's development efficiency.
🧠
Detailed Llama3 Implementation
Development of an implementation of the Llama3 language model, detailing each step of the construction, from matrix multiplication to token encoding. The use of advanced programming techniques and the meticulous approach to reading model files stand out.
🔗GitHub - naklecha/llama3-from-scratch: llama3 implementation one matrix multiplication at a time
🤖
Disintegration of the OpenAI Superalignment Team
It reveals the internal dissolution of the team responsible for AI security at OpenAI, highlighting fundamental disagreements and ethical challenges. The complexity of internal dynamics and the importance of AI security are emphasized.
🔗Why the OpenAI superalignment team in charge of AI safety imploded - Vox
📈
Gemini 1.5: A Generational Leap in AI Models
Gemini 1.5 models achieve near-perfect recall (>99%) on long-context retrieval tasks and improve the state of the art in long-document QA, long-video QA, and long-context ASR.
The Gemini 1.5 Pro and Gemini 1.5 Flash models are capable of handling extremely long contexts, remembering and reasoning about detailed information up to at least 10M tokens. This scale is unprecedented among major contemporary language models.
🔗gemini_v1_5_report.pdf (storage.googleapis.com)
📊
What's New in ChatGPT: Interactive Tables and Drive Integration
It introduces interactive tables and the ability to add files directly from Google Drive and Microsoft OneDrive. Available to ChatGPT Plus, Team and Enterprise users in the coming weeks.
🔗OpenAI on X: "We're rolling out interactive tables and charts along with the ability to add files directly from Google Drive and Microsoft OneDrive into ChatGPT. Available to ChatGPT Plus, Team, and Enterprise users over the coming weeks.https://t.co/Fu2bgMChXthttps://t.co/M9AHLx5BKr" / X
🤖
Inflection AI bets on emotional intelligence for business chatbots
Inflection AI reveals new team and plan to integrate emotional AI into enterprise bots. With less than $650 million after Microsoft deal, it focuses on empathetic chatbots for businesses.
🔗Exclusive: Inflection AI reveals new exec team and plan | VentureBeat
🎭
AI voice generates controversy with resemblance to Scarlett Johansson
Scarlett Johansson confronts OpenAI about virtual assistant's voice that mimics hers. The actress demands transparency in the use of creative work in AI. 🔗 Scarlett Johansson wants answers about ChatGPT voice that sounds like 'Her' : NPR
🤖
Artificial Intelligence and the Future of Innovation
Stephen Wolfram discusses the unpredictability of AI and its impact on science and technology. It explores the relationship between humans and machines, highlighting automation as an advancement that elevates human potential.
🔗Stephen Wolfram on Artificial Intelligence's Future (reason.com)
🚀
ChatGPT mobile application revenue reaches historic peak with launch of GPT-4o
The introduction of the GPT-4o model drove the biggest increase in revenue ever seen in OpenAI's mobile application, with emphasis on subscription to the paid service. The technical innovation of GPT-4o, which processes text, voice and video, has generated significant demand, especially on mobile devices.
🚫
Introduction of Copilot+ PCs
Introduces new Copilot+ PCs, with powerful processors and advanced AI models capable of trillion-dollar operations and real-time audio translation. Available from June 18th, starting at $999.
🔗Introducing Copilot+ PCs - The Official Microsoft Blog
🚀
Nadella's AI Era at Microsoft
Satya Nadella revitalized Microsoft, with the company reaching a market capitalization of over $3 trillion. It now faces regulatory and innovation challenges in the age of AI.
🧠
Fundamental Multimodal Early Fusion Models
Chameleon is a family of multimodal models that understand and generate images and text. It stands out for its performance in image captioning and text and image generation tasks.
🔗[2405.09818] Chameleon: Mixed-Modal Early-Fusion Foundation Models (arxiv.org)
🤖
Artificial Intelligence and the Future of Web Search
Sundar Pichai, CEO of Google, discusses the integration of AI into online search and the implications for the web ecosystem. It explores the tension between Google's futuristic vision and content creators' current concerns.
🔗Google CEO Sundar Pichai on AI-powered search and the future of the web - The Verge
🤖
Artificial Intelligence in Management
Artificial Intelligence (AI) is transforming management, requiring companies to define their role. Tools like Microsoft Copilot allow you to delegate tasks and focus on what is valued in individual contributions.
🔗AI is already changing management — companies must decide how (ft.com)
🤖
OpenAI Superalignment Team Disbanded
OpenAI disbanded the team in charge of assessing AI risks, such as “rogue” behaviors. Ilya Sutskever and Jan Leike, co-leaders, resigned, criticizing the focus on “shiny” products at the expense of security.
🔗OpenAI Scraps Team That Researched Risk of 'Rogue' AI (businessinsider.com)
🔬
Google DeepMind Border Security Framework
Introduces protocols to identify and mitigate risks from advanced AI models. Highlights the importance of balancing innovation and security in the responsible evolution of AI.
🔗Introducing the Frontier Safety Framework - Google DeepMind
🤖
'Small' Language Models Seek Big Profits
Artificial intelligence companies rely on smaller-scale language models to generate significant profits. They stand out for their efficiency and cost-benefit, without compromising generative quality.
🔗Artificial intelligence companies seek big profits from ‘small’ language models (ft.com)
🤖
The AI Revolution Is Coming
Explores the acceleration of AI innovation and its transformative impact on society. Discusses skepticism regarding AI progress and future expectations.
🔗It’s Time to Believe the AI Hype | WIRED
🤖
AI Regulation as Crony Capitalism
Explores the proposal to regulate AI as a means of crony capitalism, where business and government interests align. Discusses the potential impact on innovation and competition, warning of the risk of a government-protected AI market.
🔗AI Regulations Are Crony Capitalism in Action (reason.com)
🧠
Data Property Rights as a Solution to Runaway AI
Explores the importance of data ownership in regulating AI. Highlights the need for transparency and responsibility in the use of data.
🔗Data Ownership Rights Can Fix What Concerns Us Most about Unchecked AI | Scott McNealy — DrumWave
👓
Smart Glasses Revolution
Google's Project Astra and Meta's multimodal capabilities promise to transform the use of smart glasses. The integration of advanced AI could help people with visual impairments.
🔗Google, OpenAI, and Meta Could Revolutionize Smart Glasses (geekynews.org)
end... for today ;)
What was the news that surprised you the most? Please share your thoughts in the comments section below! 🎙️😊
For those with technical knowledge and access to the necessary resources (NVidia RTX3050+ GPU | Google Colab is a good alternative (free limited access to GPUs)):
Want to try some of these solutions locally by yourself? Discover the Open Sources solutions that are making a difference in the world of AI, the ones I have tested, parameterized and do recommend (and some original ad-hoc scripts, too): Filipe MS Bento (fmbento) / Repositories · GitHub
See all editions of this newsletter here: https://www.linkedin.com/in/filipebento/recent-activity/newsletter/
Disclaimer: The opinions and views expressed by the news, tools or services presented in this article are those of the authors and do not necessarily reflect the position or opinions of the author of this newsletter. Furthermore, the author has not added tracking codes or affiliate links.
#AINews #ArtificialIntelligence #AI #MicrosoftBuild2024 #AIPCs #ChatGPT #OpenAI #Snapchat #Canva #GoogleIO2024 #Copilot #Apple #Accounting #Retail #GenerativeAI #Gemini1.5 #InflectionAI #Management #WebSearch #SmartGlasses #Google #Meta #DeepMind #Ethics #Privacy #Security #Innovation #Technology #AIStartups #AIAccounting #AIManagement #AIWebSearch #AISecurity #AIRegulation #DataPropertyRights #AIInnovation #AIEntertainment #VoiceCloning #Chatbots #AIModels #AITools #AIApplications #AIEthics #AIChallenges #AITransformation #AIRevolution #AIIntegration #AIStrategy #AIResearch #AIDevelopment #AIAdvancements #AIPolicy #AIRegulatoryChallenges #AIModelRiskManagement #AIModelSecurity