⏐ Taklimat Pagi
Taklimat Pagi Saya
🔗 baca_penuh: pagi.hejes.my/2026/08/12
> ringkasan_ai
# 🤖 AI & Machine Learning (The Heavy Stuff)
✅ **OpenAI’s New Frontiers** [OpenAI] — OpenAI kongsi kejayaan baru dalam matematik, sains komputer teoretikal, dan macam mana dorang bina sistem suara *real-time* (GPT-Live) yang lebih natural.💡 **Kenapa Penting** — AI dah mula pecahkan masalah matematik yang lama tak selesai, bukan setakat buat puisi je.
✅ **River AI Dapat Dana Gergasi** [TechCrunch] — Startup baru umur 2 bulan, River AI (asas oleh ex-xAI), terus dapat suntikan modal $1.1 bilion untuk bina *personal agents*.💡 **Kenapa Penting** — Nilai pelaburan ni gila tinggi, tanda *personal AI agents* bakal jadi trend besar.
✅ **Gemini Capai 1 Bilion Pengguna** [TechCrunch] — App Gemini Google dah cecah 1 bilion user, dengan ramai gila guna fungsi suara dan jana 150 juta gambar sehari.💡 **Kenapa Penting** — Dominasi Google dalam pasaran chatbot makin kuat dan user makin suka interaksi suara.
✅ **Isu Ketelusan & Regulasi AI di EU** [Amanz/OpenAI] — Anthropic mula letak *invisible watermark* pada kandungan Claude, manakala OpenAI kongsi langkah keselamatan dorang untuk patuh EU AI Act.💡 **Kenapa Penting** — EU makin tegas pasal AI, jadi semua syarikat besar terpaksa ikut aturan untuk elak kena denda.
✅ **Agentic AI: Dari Alert ke Action** [Hacker Noon/arXiv] — Perbincangan pasal macam mana nak buat AI yang betul-betul boleh buat keputusan (autonomous) dan bukan sekadar bagi notifikasi atau ringkasan.💡 **Kenapa Penting** — Ini langkah seterusnya untuk AI jadi "pekerja" sebenar, bukan sekadar "pembantu".
# 💻 Tech & Development
✅ **MetaSpace & Embodied Agents** [arXiv] — Kajian baru pasal macam mana nak uji kognisi spatial untuk AI yang ada "badan" fizikal (robotik).💡 **Kenapa Penting** — Penting untuk masa depan robotik supaya AI tak langgar dinding bila bergerak.
✅ **Bluesky Makin Merosot** [TechCrunch] — User aktif Bluesky makin kurang walaupun dorang cuba kembangkan fungsi selain daripada sekadar app sosial.💡 **Kenapa Penting** — Menunjukkan betapa susahnya nak kekalkan momentum lepas *hype* pilihan raya.
✅ **HuggingFace Updates** [HuggingFace] — Pelbagai update teknikal termasuk Transformers.js, NVIDIA NeMo AutoModel, dan cara run vLLM server dengan satu command.💡 **Kenapa Penting** — Tooling untuk developer makin senang dan cepat, proses *fine-tuning* pun makin efisien.
# 📱 Gadget & Lain-lain
✅ **Kos RAM iPhone 18 Pro Melambung** [Amanz] — RAM dijangka jadi komponen paling mahal (42% kos) untuk iPhone 18 Pro sebab permintaan tinggi dari pusat data.💡 **Kenapa Penting** — Harga iPhone mungkin makin mahal sebab "perang" hardware untuk sokong AI on-device.
✅ **AI Masuk Sektor Tempatan (Malaysia)** [Amanz] — Puspakom guna AI untuk pemeriksaan kereta, dan Huawei lancarkan MatePad Air dengan fungsi AI di Malaysia.💡 **Kenapa Penting** — AI dah mula masuk dalam servis harian kita kat Malaysia, bukan sekadar dalam skrin laptop.
✅ **Scaleup Europe & ICEYE** [TechCrunch] — Dana $5.7 bilion Scaleup Europe buat pelaburan pertama dorang dalam syarikat satelit ICEYE.💡 **Kenapa Penting** — Pelaburan besar dalam teknologi satelit untuk pemantauan bumi.
---
🔥 Top Picks
1. **OpenAI’s Math & Voice Advances** — Sebab ni tunjuk AI dah mula "berfikir" tahap tinggi.
2. **River AI $1.1B Funding** — Nilai pelaburan ni memang tak masuk akal untuk startup 2 bulan.
3. **Kos RAM iPhone 18 Pro** — Sebab ni akan beri kesan terus kat poket kita nanti.
> ls -la berita/
🧠 AI/ML
15What Project Hail Mary Can Teach IT Professionals (Part 1)
Essay #1 in a 4-part series that reflects on "Project Hail Mary" (the book and movie) and mines it for lessons on leadership, persistence, and teamwork.
The Knowing-Saying Gap: When Probes See Errors that Confidence Misses
arXiv:2608.07528v1 Announce Type: new Abstract: Linear probes detect corrupted context in language models with near-perfect accuracy, yet this does not translate into reliable failure prediction. The result is a dissociation with direct implications for deployment monitoring. Across multi-hop arithmetic chains, probes that detect corruption turn out to be uninformative about final answer correctness; models forced into structured confidence formats collapse to two values with indistinguishable e
Brad Lightcap, OpenAI’s longtime COO, is leaving to ‘start something new’
One of OpenAI's longest-serving executives is headed out the door, although the longtime COO told staff that he was "excited to help you all advance the mission from a different vantage point."
HireQuotient's EasySource Delivers $60,000 Cost Savings for Stern at Home Therapy
Flagship AI Talent Platform EasySource Cuts Time-to-Hire by 45% for Healthcare Providers via 90% License Board Coverage, 20M+ Verified Profiles, and Autonomous
AITX's SARA Shows Why Most Agentic AI Never Leaves the Alert Screen
Most AI systems can detect and summarize problems but still hand decisions to humans. Explore what it takes to build AI that can reliably act.
When LLM Agents Negotiate: Private Information and Dynamic Bargaining in Supply Chains
arXiv:2608.07538v1 Announce Type: new Abstract: As LLM agents move from decision support to autonomous procurement, firms need to know whether delegated negotiators create value, divide it predictably, and avoid money-losing contracts. We study this in a canonical supply chain bargaining problem: a buyer with private demand information negotiates a quantity-payment contract with an uninformed seller. We benchmark nine LLMs from OpenAI, Google, and Alibaba against a validated Perfect Bayesian Equ
General Catalyst leads $1.1B round into 2-month-old River AI
River AI, a startup founded by xAI co-founder Igor Babuschkin, has a fascinating vision for personal agents and secured $1.1 billion out of the gate.
Shipping huggingface_hub every week with AI, open tools, and a human in the loop
$ZAMA Lists on Revolut, Bringing Blockchain Privacy Into The Mainstream
Zama, the fastest growing confidentiality protocol for onchain finance, today announced that its native token, $ZAMA, is now available for trading on Revolut, t
Kandungan Janaan Claude Akan Mempunyai Tera Air Di EU
Anthropic mula meletakkan tera air yang tidak boleh dilihat pada kandungan yang dijana oleh Claude sejak 2 Ogos lalu di negara Kesatuan Eropah (EU), Ini ialah bagi mematuhi akta kecerdasan buatan (AI) EU yang menuntut ketelusan dalam kandungan yang dijana oleh AI. Tera air ini dibenamkan ke dalam teks tetapi tidak boleh dilihat oleh mata […]
Puspakom Perkenalkan Sistem Pemeriksaan Kereta Dikuasai AI
Puspakom memperkenalkan sistem pemeriksaan kenderaan baharu berasaskan kecerdasan buatan (AI) hasil kerjasama dengan syarikat teknologi tempatan, Keymag Sdn. Bhd. Ia bertujuan meningkatkan kecekapan serta ketepatan pemeriksaan automotif sekali gus mengukuhkan peranan Puspakom sebagai pusat pemeriksaan utama dan menyokong perkembangan teknologi dalam industri automotif negara. Sistem baharu ini menggunakan AI untuk mengesan komponen serta kecacatan kenderaan […]
YouTube Perkenalkan “Practice Mode” Untuk Siaran Langsung Menegak
Ramai pengguna sering berasa gusar dengan risiko melakukan kesilapan ketika siaran langsung. Namun, keresahan itu kini dapat dikurangkan dengan pengenalan ciri baharu oleh YouTube iaitu “Practice Mode for Vertical Live“. Ciri ini direka khusus untuk mereka yang ingin berlatih mengendalikan siaran langsung dalam format menegak tanpa perlu disiarkan kepada umum. Melalui mod ini, pengguna boleh […]
Disproportionate Collapse: What Building Codes Know About Your Cascading Failures
How disproportionate collapse, a design check from building codes, maps onto cascading failures: find the load path, size the tie, name the key element.
Dynamic Coalition Formation and Communication Pricing in Skill-Based Agentic AI Systems
arXiv:2608.07532v1 Announce Type: new Abstract: Modern agentic AI systems combine multiple large language model agents with heterogeneous skills, yet most architectures either fix communication in advance or allow full broadcast. Both can be inefficient because token cost, latency, redundancy, and error propagation increase with the number of active agents and communication links. We model agent selection and communication as a cooperative game with task-conditioned net utility $U(C\mid x)=V(C\m
Pratempahan Huawei MatePad Air Baharu Dibuka Di Malaysia Dengan Deposit RM100
Huawei melancarkan MatePad Air iaitu sebuah tablet dengan rekaan ringan dan nipis yang digabungkan bersama perisian AI terkini. Ia ditawarkan sebagai pilihan untuk pengguna moden yang mahukan kemudahan dalam bentuk mudah alih. MatePad Air hadir dengan paparan OLED PaperMatte Display bersaiz 12 inci, resolusi 2800 × 1840 dan ketumpatan 275 PPI. Tubuh tablet berketebalan hanya […]
⚡ Tech/Dev
4Google’s Gemini app surges to 1 billion users
Google also shared numbers of how people are actually using the chatbot, with 63% of Gemini users talking directly to the assistant using the voice feature. Plus, Gemini now generates more than 150 million images every day, according to Google.
MetaSpace: Metamorphic Testing for Spatial Cognition in Embodied Agents
arXiv:2608.07533v1 Announce Type: new Abstract: An embodied agent is an intelligent entity that interacts with its environment through a physical body. Currently, the evaluation of embodied agents primarily relies on two paradigms: (1) manually annotated Visual Question Answering (VQA) pairs and (2) high-level task completion metrics, such as success in navigation or manipulation. The former is labor-intensive and subject to variability in annotation quality. The latter may obscure critical vuln
Bluesky’s active user base is shrinking as its focus expands beyond the app
Over a year following its post-election surge, Bluesky’s mobile app is seeing a continued decline in active users, though its remaining community is still relatively engaged.
NL2SHACL-Bench: A Benchmark Suite for Natural Language to SHACL Translation
arXiv:2608.07530v1 Announce Type: new Abstract: SHACL is a core technology for validating the conformance of RDF knowledge graphs (KGs). Yet, authoring SHACL shapes requires technical expertise that most domain experts lack. Translating natural language requirements into SHACL (NL2SHACL) would lower this barrier. However, there is no dedicated benchmark for NL2SHACL, and evaluating generated shapes requires methods beyond string comparison, as semantically equivalent shapes can differ in seriali
📌 Lain-lain
2RAM Dijangka Menyumbang 42% Kos Menghasilkan iPhone 18 Pro
Dahulu komponen paling mahal pada sebuah telefon pintar ialah panel skrin OLED atau cip pemproses. Tetapi disebabkan permintaan daripada pusat data, RAM kini melonjak menjadi komponen paling mahal dengan ini disahkan oleh pengasas Nothing, Carl Pei pada Jun lalu. Sejak Oktober 2025, harga RAM sudah meningkat lebih empat kali ganda. Kini menurut laporkan TrendForce, untuk […]
What’s Scaleup Europe, the $5.7B fund that just backed satellite company ICEYE?
Scaleup Europe, a public-private fund with a $5.7 billion target, made its first investment by backing Finnish satellite company ICEYE.