🗞 Sam Altman Predicts Rapid AI Advancements, Calls GPT-4 "Dumbest Model"

🔎 The Latest on the AI Frontier:

• OpenAI CEO Predicts Rapid AI Advancements, Calls GPT-4 "Dumbest Model" 🚀

• Tech Giants Pressure AI Engineers, Prioritizing Speed Over Safety 😓

• Apple's iPhone Sales Dip, but Services and Buyback Boost Stock 📱

• Rabbit R1's Reliance on OpenAI Exposed by Brief Outage 🐰

• Microsoft Releases Responsible AI Report, Highlighting Risk Mitigation Tools 🤖

• Microsoft Bans Police Use of Azure OpenAI for Facial Recognition 🚓

• Microsoft's Record-Breaking Renewable Energy Deal Fuels AI Growth 🌿

• Cybercheck AI Faces Legal Scrutiny Over Accuracy and Transparency ⚖️

• Generative AI Democratizes Development, Expands AI Teams' Scope 🌠

• Gaming Industry Embraces AI to Cut Costs and Enhance Experiences 🎮

• AI-Powered Second Opinion Services Could Reduce Diagnostic Errors 🩺

• Couple Saves Thousands by Using ChatGPT to Plan Their Wedding 💍

🚀 OpenAI CEO Sam Altman declares "GPT-4 is the dumbest model any of you will ever have to use again by a lot" during a Q&A session at Stanford University, highlighting the company's commitment to rapidly advancing AI capabilities. Read More

• Altman downplays concerns about the increasing costs of developing and running AI models, emphasizing the importance of providing capable tools and letting people figure out how to use them to build the future.

• Despite the success of ChatGPT, Altman describes GPT-4 as "mildly embarrassing at best" and suggests that future iterations, such as GPT-5 and GPT-6, will be significantly smarter.

• OpenAI may reveal a rival to Google's search engine as early as next week, coinciding with Google's I/O event and Apple's new iPad lineup announcement, which is rumored to feature an AI focus alongside the new M4 chip.

😓 AI engineers at top tech companies reveal intense pressure to rapidly develop and deploy AI tools, often at the expense of thorough testing and consideration of potential harms. Read More

• Engineers report working long hours and weekends on AI projects that are frequently deprioritized or shelved in favor of chasing competitors' announcements.

• Concerns about AI's impact on surveillance, climate change, and other real-world issues are often overlooked as companies prioritize speed and investor satisfaction.

• Burnout is becoming increasingly common among AI workers, with some seeking new jobs or moving away from AI departments due to the unsustainable pace of development.

📱 Apple reports a 10% drop in iPhone sales for Q2 2024, with an 8% decline in China, but beats Wall Street expectations due to increased services revenue and a $110 billion stock buyback. Read More

• Apple's slow adoption of AI compared to competitors like Google and Microsoft may have contributed to consumers holding off on purchasing new iPhones, with major AI announcements expected at WWDC in June.

• Services revenue, which includes offerings like iCloud, Apple TV+, and Apple Music, jumped 14% year-over-year, helping to offset the slowdown in hardware sales.

• Despite reports of worse than anticipated sales for the Vision Pro headset, Apple remains optimistic about the potential of spatial computing and has hinted at upcoming AI-related announcements in the weeks ahead.

🐰 Rabbit R1 users experienced a brief outage of the device's "push-to-talk" feature on Thursday, apparently caused by a concurrent OpenAI API outage, raising questions about the AI device's reliance on ChatGPT. Read More

• Rabbit CEO Jesse Lyu recently defended the Rabbit R1, stating that it is "not an Android app," but the outage suggests the device may be heavily dependent on OpenAI's technology.

• The issue was first reported in Rabbit's Discord channel at 6:12 pm ET, with users complaining of error messages and the customer support team acknowledging the problem and working on a resolution.

• OpenAI reported an API outage on its status page at 6:06 pm ET, noting "elevated error rates on fine-tuning API models and reserved capacity," and the issue was resolved approximately 45 minutes later, coinciding with the restoration of Rabbit R1's push-to-talk functionality.

🤖 Microsoft releases inaugural Responsible AI Transparency Report, highlighting 30 responsible AI tools with over 100 features to support customer-developed AI and a 16.6% increase in its responsible AI team. Read More

• Microsoft's responsible AI tools aim to identify, measure, and mitigate AI risks, with features like the Python Risk Identification Tool (PyRIT) for generative AI and evaluation tools in Azure AI Studio.

• Despite these efforts, Microsoft has faced challenges with its AI models, including incidents of inappropriate responses from its Copilot and Bing chatbots.

• The company acknowledges that responsible AI is an ongoing process and remains committed to sharing its learnings and engaging in dialogue around responsible AI practices.

🚓 Microsoft bans police departments in the United States from using its Azure OpenAI Service for facial recognition purposes, updating its code of conduct to reflect this policy. Read More

• The move comes amid growing concerns about the security, privacy, and potential for discrimination associated with the increased use of biometric technologies like facial recognition.

• The Federal Trade Commission (FTC) has expressed its commitment to addressing unfair and deceptive practices related to biometric information, while the White House has unveiled a policy requiring federal agencies to provide clear opt-out options for such technologies.

• Despite these concerns, a significant number of U.S. consumers embrace biometric authentication tools, with more than half of online shoppers using them to validate their transactions, according to a PYMNTS Intelligence and AWS collaboration.

🌿 Microsoft signs the largest corporate renewable energy purchase agreement to date, supporting the development of 10.5 gigawatts of new capacity worldwide to fuel its AI ambitions and meet its climate goals. Read More

• The deal with Brookfield Asset Management is almost eight times larger than the previous record for a single corporate power purchase agreement and could cost between $11.5 to $17 billion to build out the new capacity.

• As AI becomes more energy-intensive, Microsoft risks increasing its greenhouse gas emissions, which conflicts with its commitment to achieving carbon negativity by 2030, making the shift to renewable energy crucial.

• The new renewable energy projects will be located across the US, Europe, Latin America, Asia, and the Pacific, with Microsoft purchasing power generated in each location to match 100 percent of its electricity use with carbon-free energy by 2030.

⚖️ Cybercheck, an AI tool used in nearly 8,000 criminal cases across 40 states, faces legal challenges as defense lawyers question its accuracy, reliability, and lack of transparency. Read More

• The tool's creator, Adam Mosher, claims Cybercheck's accuracy tops 90%, but judges in New York and Ohio have barred the software's evidence due to concerns about its reliability and undisclosed methodology.

• Defense lawyers allege that Mosher lied under oath about his expertise and made false claims about the technology's use, with prosecutors in Canada stating that Mosher did not testify in cases he claimed to have been involved in.

• As the software's use continues to spread, defense lawyers are demanding access to Cybercheck's proprietary code and algorithm to ensure defendants' right to due process is protected.

🌠 The rise of generative AI has democratized AI development, creating a new class of "citizen AI developers" and expanding the scope of AI teams within organizations. Read More

• While generative AI has empowered casual users to create AI outputs, it hasn't replaced the need for highly-skilled AI specialists, such as data scientists and machine learning engineers, for more complex AI patterns.

• Modern AI teams now include AI project management roles and Trustworthy AI teams responsible for privacy, compliance, risk, ethics, governance, and intellectual property management.

• As AI becomes core to organizations, the importance of AI project managers and best practice methodologies, such as CPMAI, increases to ensure AI project success and organizational wins.

🎮 The computer games industry is embracing AI to reduce soaring production costs, save developers time, and enhance gameplay experiences, with companies like Promethean AI, Inworld, and Latitude.io leading the charge. Read More

• Promethean AI offers developers AI-powered tools to craft virtual worlds, automating repetitive tasks and providing creative feedback, while still relying on human artists to guide the process.

• Inworld's AI engine allows developers to add realism and emotional depth to game worlds and characters, with its narrative graph, developed in partnership with Xbox, using AI to help create storylines.

• Latitude.io's AI Dungeon game showcases the potential for AI to personalize gaming experiences, offering players unique, dynamic worlds and stories tailored to their choices and actions.

🩺 AI-powered second opinion services could improve clinical decision-making by addressing limitations in human cognition that contribute to diagnostic errors, which harm an estimated 795,000 Americans each year. Read More

• A recent survey found that more than half of American adults have input their health information into large language models (LLMs), with some receiving valuable insights that led to accurate diagnoses.

• The authors propose integrating AI safely into clinical workflows as a "second opinion" service to assist physicians with challenging cases and check for blind spots in diagnostic reasoning, potentially reducing the impact of cognitive biases.

• While risks such as AI biases and hallucinations must be mitigated through close human involvement, the high stakes of diagnostic errors warrant studying these technologies now, with a human expert always in the loop.

💍 Bride and groom save between $5,000 and $10,000 by using ChatGPT to plan their New York wedding, with the AI assisting in everything from finding a photographer to arranging save the date invitations. Read More

• Maria Cortese, who is set to marry her fiancé Ryan in Poughkeepsie this summer, found ChatGPT's specific responses more helpful than Google's generalized results when planning her wedding.

• According to Zola's annual First Look Report, 54% of couples getting married in 2024 said they are considering or planning to incorporate artificial intelligence tools into their wedding planning.

• While AI can assist with many aspects of wedding planning, wedding planner Katy Turich-Martin emphasizes that it cannot replace the human touch and peace of mind provided by a professional planner on the big day.

Have any feedback? Send us an email