AI agents could reshape the early career job market. Some tech leaders warn that as companies automate workflows, entry-level roles traditionally filled by recent graduates could shrink dramatically. If AI takes over routine analysis, reporting and coordination tasks, the first rung of the career ladder may start disappearing. The real question is what replaces it. If AI handles the junior work, how will the next generation gain the experience needed to become tomorrow’s leaders? https://t.co/MijPNGMAX7 @BillRMcDermott @servicenow @samantha_subin @cnbc

The best trait of 2 year olds is that if you try to do something for them, they will be like, "No! Myself!" and try to do it entirely on their own, badly. Be more like a 2 year old.

Grok 4.20 ranks #2 on 𝜏²-Bench for Telecom Agentic Tool Use on Artificial Analysis with 96.5% accuracy, outperforming Claude Opus 4.6 (max), GPT-5.4 (xhigh), and Gemini 3.1 Pro, while closing in on GLM-5 Tool calling is where the whole game is for AI agents, and this is where Grok 4.20 takes over

The problem with most people is that they have never seen real evil They are soft people who inherited good times, and assume it was always so Their concept of crime is so soft and sympathetic that they have more pity for criminals than victims. But real evil - wretched, twisted, soulless evil - is out there. If the soft people do not wake up, a small number of evil people is all that it takes to destroy vast swaths of all that is good

@juddrosenblatt Gradient-trained models are function approximators that fall within a training distribution. Outside that region the function is unconstrained. So the idea that they could infer strategies beyond their observable environment (e.g., escaping a simulation) has no basis.

@juddrosenblatt Think of it like an image through a keyhole: you can increase the resolution of what’s visible with more sampling, but you cannot reveal what’s outside the view. AI works the same, it can refine known data, but it can’t invent information it’s never seen. #keyholeprinciple #ai

@kapilansh_twt You’re absolutely right. One of the core problems with LLMs and AI agents is that they often fail silently. That isn’t a bug you can simply fix, it’s a consequence of how these systems are designed. Until the AI industry openly acknowledges this reality, we’ll likely keep being surprised by “unexplained” failures. This is the price of speculation and misaligned incentives across academia, research, and politics. For now, everyone is along for the ride, the vertigo is already paid for.

Through various initiatives, @PrincetonSPIA is informing lawmakers about the latest research on AI, and educating current and future public servants about policy challenges and innovation opportunities. https://t.co/kl4TwRAm0F

The time to learn how to think for yourself was before genAI, if you missed your chance, good luck

🥎 Saturday in Spartanburg! 🥎 🆚 Wofford 🏟️ Hope Field • Spartanburg, S.C. ⏱️ 12/2:30 p.m. 📊/💻 https://t.co/mBNujpOMYu #GoWood | #SaddleUp https://t.co/cHyxjcLwrv

🥎 Saturday in Spartanburg! 🥎 🆚 Wofford 🏟️ Hope Field • Spartanburg, S.C. ⏱️ 12/2:30 p.m. 📊/💻 https://t.co/mBNujpOMYu #GoWood | #SaddleUp https://t.co/cHyxjcLwrv

@Replit All visuals from Replit slides. This is bringinjg back excitement to write again!

🚨BREAKING: Q4 GDP growth revised down from 1.4% to 0.7%. Core PCE inflation came in at 3.1% in January. It’s only going to get worse from here. Trump’s “golden age.” https://t.co/s47GbCk8f2

If you, as a CS Prof, are wondering whether you need PhD students at all now that you have wangled a subscription to Claude Code, your lab probably had a pretty depressing vibe to begin with--and 'em students are likely better off with you hanging out with Claude.. #AIAphorisms

merjs just went from 2,400 req/s to 115,000 req/s. A 48x improvement. In one afternoon. working end to end fully on @CloudflareDev No Rust rewrite. No C extensions. Just better Zig. Here's what changed (thread) 🧵 https://t.co/pUQSVYOHmq

merjs just went from 2,400 req/s to 115,000 req/s. A 48x improvement. In one afternoon. working end to end fully on @CloudflareDev No Rust rewrite. No C extensions. Just better Zig. Here's what changed (thread) 🧵 https://t.co/pUQSVYOHmq

I made an interactive blog post about how JPEG image compression works: https://t.co/I3HU73R2ae https://t.co/USHl6t9MKG

@DaveShapi Game theory

The standard libraries already include π. But where’s the fun in that? Happy #PiDay! 🍕 https://t.co/Rt5ffxwpYO

AI is becoming part of the filmmaking process. As the Oscars approach, Hollywood is confronting how quickly AI tools are moving into areas like script development, visual effects and post-production. The discussion is no longer whether AI will influence movies, but how far its role should go. If AI becomes a creative collaborator, the film industry may soon face a deeper question. What does authorship mean when storytelling is shared with machines? https://t.co/wggMQU6LGZ @ConversationUS

✨🇨🇳This is the customized parking feature of XPeng Motors from China. You mark where you want to park, and it will automatically park there precisely. https://t.co/vdV6x9wQJR

Nearly 6 in 10 kids are now using AI to look up information. The convenience is powerful, but some experts warn that relying too heavily on AI for answers could weaken critical thinking and problem-solving skills over time. The challenge is not the technology itself, but how young minds learn to question and verify what it produces. AI may become one of the most powerful educational tools ever built. The real question is whether it will make students smarter, or simply faster at getting answers. https://t.co/sIrIajfJb2 @CNBCMakeIt

Real-time video captioning in your browser with @LiquidAI's LFM2-VL model on WebGPU. Sending every frame to a server was never going to be the answer. Imagine the bandwidth, latency and cost. Local inference. No server costs. Infinitely scalable. This is the way. https://t.co/P0vIjoBH6Y