Hextra-AI-Insight-Daily/content/en/2025-06/2025-06-30.md

linkTitle, title, weight, breadcrumbs, comments, description

linkTitle	title	weight	breadcrumbs	comments	description
06-30-Daily	06-30-Daily AI Daily	1	false	true	Daily selection of AI industry news, open source hot spots, academic frontiers and big V opinions. AI information; AI daily; AI knowledge base; AI tutorials; AI information daily; AI tools;CMU and Xiaohongshu teams have teamed up to drop an innovative tech called HoPE (Hybrid of Position Embedding) — think Hybrid Position Encoding! 🚀 They noticed that the current multimodal RoPE kinda runs out of steam when it comes to long-context semantic modeling. So, HoPE cleverly brings in zer...

AI Insights Daily 2025/6/30

AI Daily | Updated 8 AM Daily | Aggregated Data from Across the Web | Exploring Frontier Science | Unfiltered Industry Voices | Open Source Innovation Power | AI and Humanity's Future | Visit Web Version ↗️

AI Content Summary

CMU and others rolled out HoPE to supercharge VLM long video understanding, while folks from Renmin University and others fine-tuned multimodal models with MokA.
Open-source projects now feature generative AI tutorials and AI tool libraries. Gary Marcus is dropping doubts about whether pure LLMs can actually hit AGI.
AI has totally slashed the barrier for startups, sparking a shift in investment mindsets and nudging everyone to team up and seize the moment.

AI Frontier Research

1. CMU and Xiaohongshu teams have teamed up to drop an innovative tech called HoPE (Hybrid of Position Embedding) — think Hybrid Position Encoding! 🚀 They noticed that the current multimodal RoPE kinda runs out of steam when it comes to long-context semantic modeling. So, HoPE cleverly brings in zero-frequency temporal modeling and dynamic scaling strategies, which is basically like strapping "marathon running shoes" onto Visual Language Models (VLM)! This super boosts their length generalization capability for long video understanding and retrieval tasks, sending them straight to top-tier performance! 💡Seriously cool! 'Paper Link' 'Project Link'

2. Whoa! Renmin University of China and Shanghai AI Lab teams just dropped a new breakthrough: the MokA (Multimodal low-rank Adaptation) method! 🤯 They found that when fine-tuning Multimodal Large Models (MLLM), it's easy to drop the ball, missing the sweet spot between single-modal independent modeling and inter-modal interaction. But MokA, like a total balancing champ, perfectly nails this issue by cleverly mixing modality-specific A matrices, cross-modal attention mechanisms, and shared B matrices, sending the performance of multimodal tasks skyrocketing! ✨Super awesome! 'Paper Link' 'More Details'

TOP Open Source Projects

1. The "generative-ai-for-beginners" project (rocking 86,547 stars🌟) just dropped 21 courses, totally made for newbies! It's gonna walk you through mastering the skills to build generative AI. Wanna become an AI wizard? Get learning! 💪✨ 'Project Link'

2. The "system-prompts-and-models-of-ai-tools" project (already snagged 62,777 stars✨) is seriously a goldmine! It's packed with system prompts, tools, and AI models from popular AI tools and agents like Cursor and Devin. It's your go-to, all-in-one reference to help you totally crush it with AI tools! 📚💡 'Project Link'

3. The "storm" project (already sitting pretty with 24,892 stars⭐) is a total powerhouse! It's an LLM-driven knowledge management system that acts like a mini researcher, diving deep into specific topics and then whipping up full reports with proper citations. Seriously, for writing papers or doing research, it's an absolute lifesaver! 🧠✍️ 'Project Link'

Social Media Shares

1. Famed AI scholar Gary Marcus is back at it, dropping some bombshells! 🤔 He's citing papers from MIT, University of Chicago, and Harvard University, flat-out stating that pure LLMs simply cannot cook up Artificial General Intelligence (AGI)! Why not? Because they suffer from "Potemkin understanding" (aka fake understanding) and conceptual inconsistency. Basically, AI might ace its exams, but when it comes to really grasping and using concepts, it totally fumbles. The research even found that LLMs like GPT-4o, once a concept is clearly defined, see their performance nosedive 📉 when applied to real-world tasks like classification, generation, or editing. They even have conflicting representations internally for the same idea. This has totally caught the eye and sparked tests from industry heavyweights like Google DeepMind scientist Prateek Jain! Looks like AI still has a long, long way to go before hitting AGI! 💡 'More Details'
   
   

2. Tom Huang spilled the beans on Cursor's core developers' secret sauce for efficiency! 🚀 Wanna get more out of Cursor? They're showing you how to use "Parallel Agents"! By smartly combining Tab, Formed Tab, and Background Agent, you can set up a super-efficient task execution system that'll make your AI collaboration 💻 soar! Go check out how it works! 'More Details'
   
   

3. Teacher Yang Yi just dropped a really thought-provoking take: the content creation scene is totally in an "attention arbitrage window" right now 😮‍💨! He's saying folks are already using AI to "build content leverage," hinting that in the future, once AI is everywhere, human-original content is gonna get way more valuable, maybe even commanding a premium. But what's bugging him even more is the worry that AI could slowly "eat away at human spiritual culture" for super cheap — and that's way scarier than just a shift in how we create content! ✍️Food for thought... 'More Details'

4. Teacher Yang Yi reckons that in the AI era, the barrier to starting a business has basically been "smashed" by AI! 💸 The cost of putting together an MVP (Minimum Viable Product) has plummeted, making it totally possible to validate ideas super fast. His advice for aspiring entrepreneurs? Stop stressing about whether an idea will fly! Just use AI to validate an MVP in a mere 3 days, or even churn through 30 ideas in 3 months! That way, you'll pinpoint the direction truly worth pouring your heart into way faster! 🚀💡Seriously awesome! 'More Details'

5. As an AI investor, Yang Yi spilled his "secret sauce" 📈: he's not about the hard data, he's all in on qualitative metrics! He believes that figuring out if an AI startup is worth investing in boils down to five key things: the founder's grand vision for the road ahead (including PMF and scalability), how rock-solid the team's conviction is, how much efficiency AI has boosted in team management, whether the Agent has a solid feedback loop (which he says is the secret sauce for AI success!), and the scalability of the multi-agent framework. He figures stuff like user retention numbers? Those are just "by-products" that'll naturally pop up when the time is right! 🎯Talk about a sharp eye! 'More Details'

6. Someone shared a cool "new vibe" 👨‍💻 for chatting with AI to write code, and this approach is really blowing up: Instead of immediately spewing out detailed instructions, first lay out the project background and goals clearly. Then, let the AI whip up ideas based on that info, and you can align on the nitty-gritty together. This method cleverly taps into AI's super efficiency at quickly grasping context, making up for our human tendency to "run low on brain cells" when detailed planning. It seriously ramps up work efficiency in a collaborative setup! 🤝Total game-changer for programmers! 'More Details'

7. Someone's grumbling that some investors out there are still stuck using the same old mobile internet-era data metrics to size up AI projects. And what happens? They're totally failing to spot any good ones! 🤔 That's because all those traditional logics (formal, informal, even probability theory) are basically just rearview mirrors, looking at what's already happened. But the author stresses that Bayes' theorem is the real deal for future-facing decision-making, making it way better for judging investments in the AI industry! 💡Time to upgrade that investment "operating system"! 'More Details'
   
   
   
   

8. Old Ape Dashuai and his colleague Dash are straight-up saying: The arrival of AI has totally "leveled the playing field" 🏃‍♀️💨 for everyone on the planet! They reckon the massive opportunities AI brings even dwarf the internet boom from 20 years back. It lets everyone, even junior folks, bust free from resource limits and fully leverage AI to learn and create. But they also tossed out a warning: if programmers just sit on their hands and don't push forward, that "starting line" will eventually catch up and even leave you in the dust! So, actively jumping on the AI bandwagon is seriously the only way to go!

---

Listen to the Audio Version of AI Daily

🎙️ Xiaoyuzhou	📹 Douyin
Laisheng Tavern	Laisheng Intelligence Station