OpenAI is beginning to turn its attention to ‘superintelligence’
AI can now create a replica of your personality
In today’s email:
💸 OpenAI is losing money on its pricey ChatGPT Pro plan, CEO Sam Altman says
🏥 New California law prohibits using AI as basis to deny health insurance claims
🍎 Apple auto-opts everyone into having their photos analyzed by AI for landmarks
🧰 13 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.
OpenAI is beginning to turn its attention to ‘superintelligence’
OpenAI CEO Sam Altman recently shared in a blog post that the company is shifting its focus from artificial general intelligence (AGI) to the pursuit of “superintelligence.” Altman described superintelligence as tools that could dramatically accelerate scientific discovery, innovation, and prosperity. While AGI is broadly defined as systems outperforming humans at economically valuable tasks, OpenAI emphasizes the importance of equipping people with advanced tools to drive progress. Altman remains optimistic about achieving superintelligence in the near future, despite the inherent challenges and the evolving nature of AI timelines.
However, current AI systems face significant limitations, including hallucinations, mistakes, and high costs. Altman acknowledges these hurdles but expresses confidence that they can be addressed quickly. He envisions AI agents entering the workforce soon, potentially revolutionizing company output. Yet, OpenAI has previously admitted it lacks a complete solution for steering superintelligent systems safely and scaling alignment techniques to meet future demands. This raises concerns about ensuring the safe deployment of these powerful technologies.
In recent years, OpenAI has faced scrutiny for its safety practices. The company has disbanded AI safety teams and seen key researchers depart amid concerns over its commercial ambitions. Critics argue that OpenAI is prioritizing profit over safety, a claim Altman counters by pointing to the company’s track record. As OpenAI pursues superintelligence, questions remain about whether it can balance innovation with the necessary safeguards to navigate a transformative and potentially volatile technological future.
Free Notion and Unlimited AI
Thousands of startups use Notion as a connected workspace to create and share docs, take notes, manage projects, and organize knowledge—all in one place. We’re partnering with Notion to offer 6 months of new Plus plans, including unlimited Notion AI so you can try it all for free!
To redeem the Notion for Startups offer:
Submit an application using our custom link: https://ntn.so/superpowerdaily and select Superpower Daily on the partner list.
Include our partner key, STARTUP4110P37060.
Timekettle’s new earbuds offer real-time translation on calls
Timekettle has unveiled its latest innovation, the W4 Pro earbuds, at CES 2025. These earbuds bring real-time two-way translation to phone and video calls in 40 languages, with minimal user input. While there’s a slight latency of three to five seconds after speech, the system provides seamless bidirectional translations without compromising the original voice quality. The translated audio is louder than the original, and onboard AI also offers conversation summaries.
The W4 Pro earbuds build on Timekettle’s tradition of facilitating cross-lingual communication. They can still be used for face-to-face translations by sharing an earbud, similar to earlier models. Powered by the newly announced Babel OS, these earbuds promise faster translations that adapt to customizable lexicons and deliver natural human-like tonality. Babel OS will also roll out to older Timekettle devices.
Available starting Sunday for $449, the W4 Pro earbuds aim to redefine language barriers with cutting-edge technology. Whether on calls or in person, they offer a streamlined solution for effortless multilingual conversations, making them a game-changer in the translation space.
AI can now create a replica of your personality
Researchers from Stanford and Google DeepMind have developed a way to create AI-based personality replicas through a two-hour interview, capturing individuals’ values and preferences with remarkable accuracy. The study, detailed in a paper on arXiv, involved 1,000 participants from diverse backgrounds. Using these interviews, researchers created simulation agents—AI models designed to mimic the participants’ personalities. When tested against their human counterparts in personality and behavioral assessments, the agents showed an 85% similarity. These replicas aim to assist social science research by enabling studies that are costly or impractical to conduct with real humans.
Simulation agents differ from tool-based agents, which are designed to perform tasks like scheduling appointments or retrieving information. However, the insights from simulation agent research could enhance the development of both types of AI. By using qualitative interviews, researchers efficiently distilled individuals’ uniqueness into a format the AI models could replicate. This method was chosen over data-heavy approaches, like analyzing extensive emails, as it captures idiosyncratic and deeply personal information that surveys or pre-set data might miss.
While this technology opens doors for research and AI advancements, it also raises ethical concerns. Just as image-generation tools have facilitated harmful deepfakes, the ability to replicate personalities could lead to misuse, such as unauthorized representations of individuals. Furthermore, the evaluations used, including the Big Five personality traits and behavioral games, provide only a limited measure of the AI’s fidelity to the original personality. Despite these limitations, the approach demonstrates the potential for creating realistic digital twins using minimal data, making the technology more accessible and practical for various applications.
Other stuff
Sam Altman on ChatGPT’s First Two Years, Elon Musk and AI Under Trump
New California law prohibits using AI as basis to deny health insurance claims
OpenAI is losing money on its pricey ChatGPT Pro plan, CEO Sam Altman says
Killed by LLM - A memorial to the benchmarks that defined—and were defeated by—AI progress
AI’s next leap requires intimate access to your digital life
Agents, Google's whitepaper on the basics of LLM agents
Apple auto-opts everyone into having their photos analyzed by AI for landmarks
Can LLMs write better code if you keep asking them to “write better code”?
All your ChatGPT images in one place 🎉
You can now search for images, see their prompts, and download all images in one place.
Get the extension for FREEEditor's Pick
ReactAI - Open source free AI react components builder
Lambda - Your AI-Powered Investment Copilot
Symphony - An OS where AI controls the keyboard and mouse
Infinite Convo - The never-ending AI talk-show
OpenBolt - Free AI-powered full-stack project builder
Cure AI - Your companion for evidence-based research
Dreamina - All-in-one AI creative suite for all your artistic work
GitDiagram - Instantly visualize any codebase as an interactive diagram
Supascout - Analyze App Store reviews to gain actionable insights
Monkt - Transform files and web pages into AI-ready Markdown or JSON
Lecca.io - AI agents & automation
FirstVersion - Rebuilding the first versions of great products with AI
EasyLang AI - Learn any language with AI in a fun, easy, personalized way
Unclassified 🌀
WFH Team - Work from anywhere in the world
Keep your meeting data safe
Meet Fellow, the only AI Meeting Assistant built with security and privacy at its core.
Record, transcribe, and summarize meetings with peace of mind that your meeting data is secure.
Claim 90 days of unlimited AI notes today.
Get started free
