Greetings Qrew,

Whether you're diving into AI for work, side projects, or just out of curiosity — this issue of The Creators Hive brings you practical insights, hands-on learning, and tools you can actually use.

1. Jony Ive details OpenAI’s hardware vision
2. AI researcher leaves Anthropic over anti-China stance
3. Samsung researcher’s tiny model out-reasons giants
4. Google’s unified workplace AI platform
5. Public doubles AI use — but rejects it in news
6. Tool Recommendation of the Day: natively.dev

---

Jony Ive details OpenAI’s hardware vision

Former Apple design chief Jony Ive offered new insight into his partnership with OpenAI during an exclusive Dev Day session alongside Sam Altman, outlining plans for a family of AI devices designed to heal humanity’s fractured relationship with technology.

Key details:

• Ive described today’s digital experience as an “uncomfortable relationship” with tech, saying AI devices should make us “happy, fulfilled, peaceful, less anxious, and less disconnected.”
• His team has already developed 15–20 product concepts for a “family of devices,” following OpenAI’s $6.5B acquisition of his design studio, io, earlier this year.
• Ive said it’s “absurd” to think AI could be delivered through legacy devices, while Altman emphasized that new hardware must have a “truly compelling reason” to exist.
• Altman told The Rundown the effort will “require patience” to “develop a totally new way to use a computer.”

Why it matters: Ive and Altman’s comments suggest a move away from attention-driven tech toward emotionally intelligent devices that prioritize human well-being. The vision is ambitious — but with Altman urging patience, this could be a slow-burn revolution in how we interact with AI.

---

AI researcher leaves Anthropic over anti-China stance

Physicist-turned-AI researcher Yao Shunyu has left Anthropic for Google DeepMind, citing discomfort with the startup’s classification of China as an “adversarial nation” as one of the key reasons behind his departure.

Key details:

• Yao worked on Claude 3.7 Sonnet and Claude 4 during his year at Anthropic before resigning in mid-September.
• In a blog post, he attributed 40% of his decision to Anthropic’s internal policy that bars subsidiaries from countries deemed “adversarial,” including China.
• He also referenced other “undisclosed internal matters,” noting that while his tenure was professionally rewarding, “it is better without you.”
• Yao has joined DeepMind as a senior research scientist on the Gemini team, contributing to Google’s frontier foundation model efforts.

Why it matters: This highlights how geopolitical tensions in AI aren’t just shaping national strategies but also influencing individual career moves. As competition for top AI talent heats up, company policies on international collaboration and access could become as decisive as pay or compute in attracting and retaining researchers.

---

Samsung researcher’s tiny model out-reasons giants

Samsung researcher Alexia Jolicoeur-Martineau has unveiled the Tiny Recursion Model (TRM) — a 7M parameter AI that outperformed DeepSeek R1 and Gemini 2.5 Pro on complex reasoning benchmarks by using an internal self-improvement loop that drafts, critiques, and refines its own answers.

Key details:

• TRM achieved 45% on ARC-AGI-1 and 8% on ARC-AGI-2, outperforming models thousands of times larger in parameter count.
• Instead of predicting token by token, it iteratively drafts and refines complete solutions through up to 16 internal reasoning cycles.
• Within each cycle, the model maintains a scratchpad for self-critique and logic revision before updating its answer draft.
• The model’s strengths are concentrated in abstract reasoning and pattern recognition, meaning its results don’t yet generalize across all reasoning tasks.

Why it matters: As the AI arms race pushes toward billion-parameter behemoths, TRM — like Sapient’s HRM — demonstrates that architectural ingenuity can rival brute-force scale. If such recursive systems can generalize, they may redefine what efficiency and intelligence look like in the next wave of model design.

---

Google’s unified workplace AI platform

Google has launched Gemini Enterprise, a unified AI platform that merges its workplace tools into a single system — enabling employees to build, deploy, and manage AI agents without any coding experience.

The details:

• Combines no-code agent builders with prebuilt assistants for tasks like research, customer support, and software development.
• Integrates securely with company data across multiple platforms, and features an agent marketplace offering thousands of partner-built solutions.
• Pricing starts at $30/month per user for the Enterprise tier, while a $21/month Business plan provides reduced cloud storage and limited features.

Why it matters: By merging creation, deployment, and management into one seamless ecosystem, Google is positioning Gemini as the operating layer for enterprise AI workflows. With Amazon’s Quick Suite launching the same day, the competition is no longer about whose model is smartest — but whose platform removes the most friction between people and intelligent systems.

---

Public doubles AI use — but rejects it in news

A new Reuters Institute survey across six countries shows that while weekly AI use has nearly doubled since last year — mainly for research and information seeking — public trust in AI-generated news has sharply declined.

The details:

• 24% now use AI for research and questions, surpassing text, image, and code generation (21%).
• ChatGPT remains the most-used platform, while Google and Microsoft’s AI search tools expose over half (54%) of users to AI-generated summaries.
• Just 12% of respondents are comfortable with fully AI-produced news, while 62% prefer entirely human journalism — widening the trust gap from 2024.
• Respondents rated healthcare, science, and search as positive AI applications, but were negative toward AI in news and politics.

Why it matters: The findings reveal a paradox — people embrace AI as a personal assistant yet reject it as a public information source. For publishers, that means walking a fine line between leveraging AI tools for efficiency and maintaining the human credibility their audiences demand.

---

Tool Recommendation of the Day: natively.dev

Website: natively.dev

What makes it unique

Natively is a no-code, AI-powered mobile app builder that turns your app ideas into production-ready iOS and Android applications without writing a single line of code.
You simply describe your idea in plain English, and the platform automatically generates the native code, user interface, and logic. Unlike webview-based tools, the apps are fully native and have access to device features and APIs.

Key Features

• Generate apps instantly from text prompts
• Real-time preview and testing
• Fully native performance and integrations
• Export complete source code with full ownership
• Seamless deployment to app stores
• Integrate custom features and plugins

Use case magic

• Quickly prototype and launch mobile apps
• Ideal for startups and solo founders without developers
• Convert web apps into native mobile experiences
• Build internal tools or MVPs efficiently

Why it stands out

Most app builders offer no-code convenience at the cost of flexibility. Natively bridges that gap by creating real native code you can own, edit, and scale. It combines AI-powered simplicity with developer-level control.

---

Masterpieces from the Hive 🎨

Here’s what our community has been creating this week using the Hive Wall:

32 North Gould Street, Sheridan, Wyoming 82801
​Unsubscribe · Preferences