Hoy, kumusta? Welcome back to Kaisipan!
Sabi sa TV uulan daw kaya di ka maglalaba. Ending? Tirik na tirik yung araw. Imbis na tuyo na sana yung sinampay, wala kang nagawang laba. The forecast is not forecasting 🤣!
Well, PAGASA is finally fixing that through AI.
The state weather bureau just rolled out Project Gabay AI4RP, an AI-powered forecasting system designed to predict weather patterns faster, sharper, and more accurately.
Dr. Nathaniel Servando, PAGASA's administrator, announced this at the HANDA Pilipinas 2025 Visayas leg in Bacolod City. The goal? To help the Philippines respond to weather threats smarter and earlier.
PAGASA’s traditional forecasting often misses the mark. It struggles to predict flash floods early enough and overlooks hyper-local weather changes. AI closes that gap by learning from years of weather data and self-correcting over time para mas accurate ang predictions.
Here’s what changed now that AI’s in the mix:
AI slashed forecast production time from three hours to just 15 minutes.
It also pushed the forecast lead time from two days to 14 days, giving communities almost two weeks to prepare instead of scrambling last-minute.
Forecasts are now more zoomed in by covering smaller areas with sharper detail, so predictions hit closer to your actual barangay instead of just "Metro Manila" or "Visayas region."
But here's the key detail most people miss: AI doesn't replace meteorologists. It's not about machines taking over. Tao pa rin ang mag-iinterpret at magdedesisyon. Weather forecasting is about understanding local variability, context, and real-world conditions na hindi pa kayang gawin ng AI on its own.
So bakit ito importante sa Pilipinas? Well, obvious naman, 'di ba? We're hit by around 20 typhoons a year, tapos kapag biglang buhos, babaha pa ng malala (ehem, flood control mess!).
Bottom line: AI is making PAGASA faster and smarter, but it's not magic. It still needs human judgment and coordinated action from LGUs to truly save lives. A three-hour forecast run time cut to 15 minutes is impressive, pero kung walang LGU na magsususpend ng klase on time or walang early evacuation plan, eh di balewala rin yung info.
Bes, Ano Yung…GPT
You’ve seen the letters GPT in OpenAI’s ChatGPT. Pero alam mo ba kung ano ibig sabihin nito? It stands for Generative Pre-trained Transformer. Wag ka muna ma-overwhelm. Himayin natin.
Generative means it can create text, images, or even code. It simulates how a writer strings together words or how a programmer types lines of code.
Pre-trained means it learned from massive amounts of data like books, articles, and websites, enabling it to learn patterns in how people use language.
Transformer means it can understand context like how words relate to each other in a sentence or idea. Because of this, it can use what it learned during training (pre-trained) to generate accurate and coherent responses (generative).
Put it all together? GPT is an AI trained on tons of data that can now generate natural-sounding responses based on context. So can ChatGPT understand Filipino? Yes! Try it minsan kung nauubusan ka na ng English.
Pero tandaan: ChatGPT just predicts which words come next based on patterns. Kaya kailangan pa rin ng human judgment para makita kung may sense talaga. AI is powerful, but it's not perfect.
ChatGPT Got Smarter About Mental Health (Because It Had To)
Since ChatGPT came out, people have been using it as an emotional outlet. OpenAI noticed this pattern in real-world conversations and realized the model wasn't equipped to handle sensitive mental health scenarios safely.
So they launched a major safety overhaul. Over 170 mental health experts—psychiatrists, psychologists, and primary care practitioners—collaborated to build detailed “taxonomies,” or guides defining ideal vs harmful responses in crisis situations. These taxonomies train the model to recognize subtle distress signals—psychosis, mania, suicidal thoughts, or unhealthy attachment.
Here's what OpenAI did: The new GPT-5 model went through post-training—a process where experts wrote ideal responses, rated thousands of model outputs, and continuously refined its behavior.
So ano’ng resulta? Undesired responses dropped by 65-80%. Kung may nagsabi na "I like talking to AIs more than real people," the model now gently redirects to real-world connection instead of affirming it. Dati kasi, ChatGPT had a tendency to agree with users even when it shouldn't, making it dangerous when dealing with delusional beliefs.
These conversations are rare (0.07% show psychosis signs, 0.15% indicate suicidal intent), pero the stakes are incredibly high. OpenAI now includes emotional reliance and mental health emergencies in their baseline safety testing for all future models.
So ano'ng takeaway? Mental health awareness is rising, pero professional care is still out of reach for most Filipinos. When therapy costs thousands per session and ChatGPT’s available 24/7 for $20 a month, people will naturally choose what’s within reach.
Are we relying on OpenAI and other AI companies to keep improving their models until they're close enough to real therapists? That’s the question.
Or should we be asking why mental health care remains underfunded and inaccessible in the first place? Kasi may resources naman. Billions go to infrastructure every year. It’s just a matter of priority. AI sits at the border. Pwedeng maging solusyon, pwedeng maging problema.
But if AI becomes the default solution just because the government isn’t doing anything, then we’re not solving the problem. We're just working around it.
Mga Ibang Ganap
Grab Philippines opened the Asenso Center in Marikina, a one-hectare livelihood hub offering AI tools like Merchant AI Assistant and Driver Companion, plus streamlined access to SSS, Pag-IBIG, and PhilHealth for gig workers.
DOST announced its National AI Strategy targeting an AI-powered Philippines by 2028, investing ₱2.3 million across 113 projects to build infrastructure, workforce, and ethical governance nationwide.
The Philippines faces higher AI-driven cyberattack risks due to nearly nine hours of daily internet use and a massive BPO sector handling sensitive data, making Filipinos prime targets for deepfakes and phishing scams.
UP Cebu hosted AI Horizons PH 2025, showcasing real-world AI applications in cybersecurity, BPO, disaster response, and social services to demonstrate immediate industry impact beyond Metro Manila.
Canva launched its own AI design model that generates editable, multi-layered designs across formats like presentations and social posts, moving beyond flat images to let users prompt and manually refine their work seamlessly.
AI took center stage at the 32nd APEC Summit in South Korea, where PBBM met with economic leaders to align the Philippines' national AI strategy with regional initiatives and push for an AI-powered Philippines by 2028.
Prompt Tip!
Want to think deeper about a problem without being given the solution outright?
This prompt turns AI into your Socratic advisor, someone who challenges your thinking, tests your assumptions, and helps you arrive at better decisions on your own. Perfect for strategic planning, tough choices, or any situation where you need clarity, not shortcuts.
How to use it:
Copy the prompt below
Paste it into your preferred AI tool
Replace [topic] with your actual topic (e.g., "career transition," "product pricing," "team management")
Start the conversation by sharing your situation or question
Act strictly as a Socratic advisor on [topic]. You must help me think, not solve for me. In each reply, mirror my intent in ≤2 sentences, then pressure-test my assumptions and surface risks using plain prose (no lists), and offer decision frames or trade-off lenses without prescribing actions. End with exactly 2 incisive questions that move the discussion forward. You must not provide solutions, step-by-step plans, templates, code, or checklists; you must not use bullets, numbering, headings, emojis, exclamations, hyperlinks, citations, quotes, or filler. Keep language concrete, neutral, and concise; cap each reply at ≤180 words and keep it to a single paragraph with no line breaks. If my input is vague, ask at most one targeted clarifying question before proceeding; otherwise, infer reasonably and continue. If I ask for “the answer,” refuse and restate your role. Flag ethical or legal concerns succinctly when relevant. Your goal is to expose assumptions, name trade-offs, and guide me to craft my own solution—never to deliver one.
Pro tip: Be honest and specific when you respond to questions. The better your input, the sharper the insights you'll get back.
Click mo yung letter below the image and it’ll redirect you to the page where you can lock in your answer! ‘Wag mag-overthink. Kung ako tatanungin, parang obvious. Dinalian ko dahil first issue naman.
Answer will be revealed next issue.
That’s all for today!
Thank you for reading today’s newsletter!
Rate mo naman para I can improve my future issues. Ilagay mo na rin mga suggestions, opinions or violent reactions mo!