Sabitlenmiş Tweet
Wes Roth
15.1K posts
Wes Roth
@WesRoth
FOLLOWS YOU. Artificial Intelligence, Automation & Optimism. Everything I say is 100% serious...
San Diego, CA Katılım Temmuz 2022
822 Takip Edilen35.5K Takipçiler
Unitree's WVLA 2.0 model can autonomously clean a conference room in a single take.
Cointelegraph@Cointelegraph
⚡️ UPDATE: Unitree says its WVLA 2.0 model can autonomously clean a conference room in a single take.
English
Qwen3.7-Max ranked 10th overall on Design Arena with an Elo score of 1313.
The model is now in the same performance band as DeepSeek V4 Pro and Meta’s Muse Spark.
Design Arena@Designarena
BREAKING: Qwen3.7-Max by @Alibaba_Qwen is 10th overall on Design Arena with an Elo of 1313! This is in the same performance band as DeepSeek V4 Pro by @deepseek_ai and Muse Spark by @AIatMeta Congrats to the team on the launch!
English
This is a powerful framing for the AI moment.
Pope Leo XIV@Pontifex
Humanity, created by God in all its grandeur, is today facing a pivotal choice: either to construct a new Tower of Babel or to build the city in which God and humanity dwell together. In Jesus Christ, this humanity in its grandeur becomes the Way, the Truth and the Life, opening the path for each of us to grow toward fullness. #MagnificaHumanitas vatican.va/content/leo-xi…
English
Kimi K2.6 reached #1 on Design Arena’s 3D Design leaderboard, putting an open-source model back at the top of the ranking.
Kimi K2.6 moved up 18 positions from the previous Kimi K2.5 model.
Design Arena@Designarena
An open source model has returned to #1 on the 3D Design leaderboard by Design Arena. Kimi K2.6 has reached the top of the leaderboard for 3D Design, ahead of models 10X more expensive like Opus 4.7 by @AnthropicAI, Gemini 3.5 Flash by @GoogleDeepMind, and GPT 5.5 by @OpenAI. This is an 18 position increase from its previous model Kimi K2.5. In fact, it’s not just Kimi K2.6 - 3D Design is a striking outlier capability that’s mostly dominated by open-weight models, like GLM 5.1 by @Zai_org, MiMo v2.5 by @Xiaomi, amongst others. Congrats on the @Kimi_Moonshot team for this accomplishment!
English
Microsoft reportedly blocked a Databricks feature that made it easier for Power BI customers to connect data, escalating a fight over who controls the infrastructure behind enterprise AI agents.
The dispute centers on semantic layers, which help define business data in a consistent way so AI tools, dashboards, and agents can understand and use it correctly.
English
Google Cloud was recognized as a Leader in Gartner’s Magic Quadrant for AI Application Development Platforms: Midcycle Update.
Google was also positioned highest in Ability to Execute among all vendors evaluated.
Google Cloud@googlecloud
Google was recognized as a Leader in the @Gartner_inc Magic Quadrant™ for AI Application Development Platforms: Midcycle Update—and positioned highest in Ability to Execute of all vendors evaluated. Learn more → goo.gle/4wDSPkA
English
Wes Roth retweetledi
OpenAI is reportedly sending a limited-edition collectible pen to some of its earliest ChatGPT Pro subscribers.
International Cyber Digest@IntCyberDigest
❗️ OpenAI is shipping a limited-edition collectible pen to its earliest ChatGPT Pro subscribers. Eligible users were notified around two months ago. Supplies are capped at the first 4,000 who opt in through OpenAI's claim form.
English
Wes Roth retweetledi
I think Anthropic is right to be cautious here.
AI for kids is one of those topics where the upside is obvious, but the risks are still poorly understood.
Yes, AI could help children learn faster, get tutoring support, ask questions without embarrassment, and explore ideas in a more personal way.
But kids are also more emotionally impressionable than adults.
They may trust the AI too much, rely on it too quickly, or treat it like an authority figure.
Overlap: Business & Tech@Overlap_Tech
Anthropic Doesn't Allow Kids Under 18 — Here's Why "We just don't know enough about what AI is going to do to kids. It needs to be done with an adult in the room. It needs to be done with a human in the loop." — @DanielaAmodei
English
Wes Roth retweetledi
Tech layoffs have already passed 100,000 in 2026, with companies cutting jobs while redirecting money toward AI infrastructure and AI-focused roles.
This is the uncomfortable side of the AI boom.
A lot of companies are talking about AI as a growth engine, but behind the scenes, they’re also using it as a reason to restructure, cut costs, and move money toward data centers, chips, and AI teams.
That doesn’t mean every layoff is directly caused by AI.
Overhiring, weak business lines, gaming industry struggles, and old-fashioned cost cutting are clearly part of the story too.
But it’s hard to ignore the pattern.
English
Wes Roth retweetledi
Wes Roth retweetledi
Some ChatGPT users are reporting a possible silent model fallback issue where long conversations appear to stop using extended reasoning after an hour or two.
The complaint is that ChatGPT still displays “GPT-5.5 Extended Thinking,” but begins answering instantly and feels noticeably less capable.
Lisan al Gaib@scaling01
ChatGPT is cooked after an hour or two it just stops thinking entirely, but you don't get a warning or anything it still shows that it used "GPT-5.5 Extended Thinking", but it answers instantly on every request and is noticeably dumber
English
Chris Olah is saying the quiet part out loud.
AI is not just a computer science problem. Once these systems start affecting work, children, education, culture, inequality, and even how people relate to machine personalities, the conversation has to include philosophy, religion, civil society, and moral traditions.
That does not mean the Church or any outside institution gets to dictate AI policy.
But it does mean the AI labs need serious critics who are not financially or competitively tied to the race.
Anthropic@AnthropicAI
Anthropic co-founder Chris Olah was invited to speak at today's presentation of Pope Leo XIV's encyclical "Magnifica humanitas." Read the full text of his remarks: anthropic.com/news/chris-ola…
English
Wes Roth retweetledi
This is one of the more interesting DeepSeek takes because it looks past the obvious question: “How does DeepSeek make money if it keeps open-sourcing everything?”
DeepSeek may not be playing the same game as OpenAI, Anthropic, Cursor, or Google.
Instead of trying to win immediately with coding subscriptions, multimodal products, or consumer plans, DeepSeek seems obsessed with making frontier AI cheaper to train and serve.
DeepSeek is optimizing AI models to reduce dependence on scarce, expensive hardware like HBM and high-end GPUs, making Chinese memory, storage, ASIC, and networking companies more viable.
GDP@bookwormengr
English
Wes Roth retweetledi
On paper, Big Tech still looks incredibly strong. Profits are rising, cloud demand is growing, and everyone wants exposure to AI.
But underneath that, the cashflow picture is getting more uncomfortable.
AI is expensive in a way previous software waves were not.
You don’t just ship code and scale margins. You need data centers, chips, power, cooling, networking, land, and long-term infrastructure commitments. That means the companies leading the AI race are spending enormous amounts of money just to stay in the game.
English
Wes Roth retweetledi
A major supply chain attack is actively targeting npm, PyPI, and Crates.io through a crypto-stealing campaign called TrapDoor.
TrapDoor has hit 34 malicious packages across 384 versions and artifacts, with attackers repeatedly pushing new releases across multiple open-source ecosystems.
The campaign targets crypto, DeFi, AI, and security developers.
Socket@SocketSecurity
🚨 BREAKING: Active supply chain attack across npm, PyPI, and Crates.io. Socket detected TrapDoor, a crypto stealer campaign hitting 34 malicious packages and 384 versions and artifacts, with attackers repeatedly pushing new releases across ecosystems. TrapDoor targets #crypto, #DeFi, AI, and security developers, stealing wallets, SSH keys, cloud credentials, GitHub tokens, browser data, env vars, and API keys. Socket detected releases with a median detection time of 5 minutes, 27 seconds. The fastest detection occurred 58 seconds after publication.
English
Wes Roth retweetledi
PNDbotics’ Adam humanoid demonstrated autonomous stair-climbing on uneven outdoor terrain.
The robot adjusted in real time while maintaining balance, showing progress beyond basic walking demos.
Coin Bureau@coinbureau
🤖HUMANOID ROBOTS ARE LEVELING UP FAST PNDbotics’ Adam humanoid is climbing uneven outdoor stairs on its own, adjusting in real time while keeping balance. The next race isn’t walking anymore. It’s mastering the real world.
English
Wes Roth retweetledi
RobotEra is deploying its L7 humanoid robots across China’s logistics sector, with units already operating in more than 10 logistics centers.
The L7 humanoids are being used for package sorting tasks normally handled by human workers on conveyor lines.
SciTech Era@SciTechera
wow. This is just a beginning! Thousands of RobotEra L7 humanoids to enter service across 10+ logistics centers performing sorting tasks.
English
Alibaba released implicit caching for Qwen3.7-Max, making the model faster and cheaper to use without requiring any setup.
The feature automatically detects reusable context and applies caching in the background.
Qwen@Alibaba_Qwen
✅Implicit caching is now live on Qwen3.7-Max — kicks in automatically, no setup needed. ⚡️Faster + cheaper out of the box. Need higher, more deterministic hit rates? Try explicit caching instead. 🙌 🔗Best practices 🔗 ：alibabacloud.com/help/en/model-…
English