Future SliceThe future, one slice at a time.
English
English简体中文日本語
SILICON TIDE / INDUSTRY VOICES

Silicon Tide

Original X posts in a masonry board. Native cards show the author’s words and media; our multilingual context is available underneath.

Selection checked:

Original cards load automatically as you scroll. You can switch to editorial summaries; unavailable posts keep their original links.

20 accounts · 22 posts

Products@bot

Loading X card…

Editorial summary

Grok Bot adds X monitoring

The official Grok Bot account announces support for searching, reading, and monitoring X. The post describes a product capability; it does not establish coverage limits, reliability, or access for every plan.

Open original post
Products@sama

Loading X card…

Editorial summary

ChatGPT moves toward interactive answers

Altman highlights ChatGPT’s ability to generate a custom interface, sharing OpenAI’s GPT-6 and Intelligent UI announcement. The company describes interactive answers for everyday questions and more visual explanations of complex topics.

Open original post
Products@gdb

Loading X card…

Editorial summary

ChatGPT answers become interactive interfaces

Brockman introduces Intelligent UI in ChatGPT: answers can include interactive user interfaces. His post shares OpenAI’s rollout announcement and a video demonstration.

Open original post
AI@AravSrinivas

Loading X card…

Editorial summary

Perplexity shares multimodal embedding weights

Srinivas announces open weights for pplx-embed-v2-late, with 9B and 0.6B models sharing a text-and-image embedding space. He describes larger-model indexing and on-device queries with the smaller model.

Open original post
Strategy@elonmusk

Loading X card…

Editorial summary

Grok Bot’s proposed multi-model direction

Musk says Grok Bot will choose the backend model best suited to each task, including third-party APIs. His post describes a direction going forward, without confirming deployment timing, pricing, or current integrations.

Open original post
AI@sama

Loading X card…

Editorial summary

Altman points to AI progress in mathematics

Altman links to OpenAI’s mathematics research update and presents it as a new stage of discovery. This is a research signal rather than an announcement that the underlying model is publicly available.

Open original post
AI@ClementDelangue

Loading X card…

Editorial summary

Coding-agent harnesses become RL environments

Delangue describes turning existing coding-agent harnesses into reinforcement-learning environments without modifying the harnesses or training code. He says the tooling supports open models and task sets, with the implementation open source.

Open original post
Strategy@levie

Loading X card…

Editorial summary

The technical work that AI deployment creates

Levie argues that deploying AI creates work for applied-AI engineers, forward-deployed engineers and service firms. He focuses on the surrounding work needed to make AI useful in companies.

Open original post
Strategy@rauchg

Loading X card…

Editorial summary

Writing for people versus writing for agents

Rauch describes keeping a new project’s README handwritten for human readers, while using AI-written internal documentation for agents. He treats public writing as communication between people.

Open original post
AI@karpathy

Loading X card…

Editorial summary

Make model output easier to understand

Karpathy expects interpreting model output to take more of our time and suggests trying ASD-STE100 controlled language for clearer explanations. This is a personal workflow suggestion, not a measured comprehension result.

Open original post
Products@OpenAI

Loading X card…

Editorial summary

OpenAI introduces dots

OpenAI’s official account introduces dots, describing always-on agents powered by GPT-6 Astra. This is a company product announcement; the short post does not detail operational limits, permissions, or pricing.

Open original post
Products@drfeifei

Loading X card…

Editorial summary

Li shares the next chapter for World Labs

Li links her article about World Labs joining AMD. Its preview frames this as a new chapter for the team. This is her first-party announcement; the card does not claim a completed technical integration.

Open original post
Products@satyanadella

Loading X card…

Editorial summary

Copilot’s direction as a work platform

Nadella outlines Copilot’s direction across models, devices, and tasks, highlighting a proactive, long-running enterprise agent called Autopilot. The post announces a product update rather than an independent performance comparison.

Open original post
Hardware@sundarpichai

Loading X card…

Editorial summary

Google plans an orbital TPU test

Pichai describes Project Suncatcher’s prototype satellite, developed with Planet, joining SpaceX’s Transporter-18 mission to test TPUs in space. The post frames an experiment, without reporting successful orbital operation.

Open original post
AI@fchollet

Loading X card…

Editorial summary

Delegate code while keeping understanding

Chollet argues that delegating coding still requires engineers to retain understanding. He suggests explicit artifacts and new workflows when code-centric methods no longer provide that understanding. The quoted workplace account is unverified.

Open original post
Strategy@ylecun

Loading X card…

Editorial summary

LeCun distinguishes LLM criticism from stopping research

LeCun says he has not called for LLM research to stop and points to his support for earlier open models at Meta. This card records his stated position, rather than adjudicating the broader debate.

Open original post
Strategy@mustafasuleyman

Loading X card…

Editorial summary

Suleyman challenges the idea of model welfare

Suleyman argues against treating models as conscious moral patients. His linked essay warns that anthropomorphic training may complicate alignment and calls for public debate; these are his views, not an established scientific conclusion.

Open original post
Strategy@demishassabis

Loading X card…

Editorial summary

Hassabis supports a coordinated frontier-AI approach

Hassabis welcomes the direction of Dario Amodei’s essay while noting that details still need work. He connects that discussion to his earlier proposal for an industry-wide standards body for frontier AI.

Open original post
Products@AndrewYNg

Loading X card…

Editorial summary

AI engineers should help shape the product

Ng argues that AI engineering should influence what gets built and drive the development loop, rather than only implement requirements. His linked skills map frames product judgment as part of engineering work.

Open original post
Products@sundarpichai

Loading X card…

Editorial summary

Gemini voice reaches everyday Workspace tasks

Pichai announces conversational Gemini capabilities for Gmail search, Keep organization, and creating Docs, rolling out to Google AI subscribers. He specifically highlights Docs Live; the post does not say every subscription has identical access.

Open original post
Strategy@JensenHuang

Loading X card…

Editorial summary

Huang emphasizes open models and local control

Huang highlights NVIDIA and Hugging Face while arguing that open models support cybersecurity, innovation, and sovereignty. He stresses opportunities for developers, universities, industries, and countries to customize AI; these are his stated benefits.

Open original post
Hardware@DrJimFan

Loading X card…

Editorial summary

ENPIRE brings coding agents to robot experiments

Fan introduces ENPIRE, describing eight Codex agents working with a robot fleet and allocated compute. This June announcement is retained as background on physical AI; its claims should be read alongside the project’s research.

Open original post