Browse Comments — Raw (as collected)

Interesting Medicine That's such a great extension of the idea — it really does show up everywhere. In health, in business, in relationships — the inputs matter, but the process matters just as much. A lead list of 10,000 contacts prepared poorly gets worse results than 500 contacts worked with precision and context. Same data. Different preparation. Completely different pipeline. Love the content you're putting out — the way you make evidence-based health insight accessible is genuinely rare. Following for more. 🙏

Nicole Kaminski 🧐🙃

Wow. maybe not the correct place but Can i get a referral , I am an AI Research Engineer. I can send you the details if you feel okey with it.

Jofre Ayala, LUTCF®, CPIA® what lol not liking tomatoes? Or my body being allergic to garlic, carrots, mushrooms etc.?

Good morning Nicole. It's the sincerity behind your words that makes your point meaningful.

So the idea of going to school and coming out to look for a job is global. The education system needs a total overhaul. Your education is no good if you cannot create employment, this may not be true for professional courses like Medecine and Law...however the vast majority of students should be equipped in school to create employment

Terence Tao point about having more room to explore ambitious ideas resonates. Sometimes the biggest limitation in research isn't a lack of ideas, but the time required to pursue them.

This is quite valuable Abhishek Veeramalla One challenge with learning AI today is not the lack of resources but the abundance of disconnected resources. You learn one framework here, another tutorial there, and still struggle to understand how everything works together in a real-world application. Resources that bridge the gap between theory and implementation are always worth paying attention to.

The biggest AI risk for many organizations may not be the technology itself. It may be adopting AI faster than the organization can operationally, ethically, and culturally absorb it. Every AI workflow sits on top of real infrastructure: energy, data centers, compute, cost, governance, privacy, security, and human decision-making. Yet many companies are still treating AI like a productivity plug-in. That gap matters. When AI starts changing how work is designed, how decisions are made, how employees learn, and how leaders measure performance, it becomes a People Operations issue as much as a technology issue. This is why HR leaders need to understand AI beyond adoption campaigns. We need to understand systems, workflows, governance, workforce capability, and the human consequences of scale. AI may run on infrastructure, but responsible adoption runs on leadership.

I'm surprised Laura the AI director is called it chatGPT when it's actually open ai. Graduates will need to adapt. Obviously jobs will decrease not increase.

Gaute Hesthagen Terje Boye greit å vite med tanke på prosjekt X.

I felt like throwing up after only a few seconds...

Ar. Yash Arora great question. It will be more challenging for longer clips. You almost have to visualize it and calculate in your mind the duration of the flight path before generating the videos

The idea of AI reducing cognitive friction so researchers can chase the "too ambitious" paths is one of the most compelling use cases for this technology. When someone like Terence Tao says it expands what he can attempt that's not a small statement. The best breakthroughs often came from questions nobody had time to fully pursue.

Good

Jose Luis Flores® ☁ you are welcome

Ankit Vishwakarma you are welcome

Sherry Horowitz same here! I also jump between Nano Banana and GPT-Image-2 depending on the use case. I’m glad we have multiple capable image and video models to choose from

SYD HOSSEINI the aspect ratio depends on what AI model you choose. The most standard ones that all major AI models support are 16:9 and 9:16. 1080p to 4K. But at the moment I’m only comfortable with the qualities from 720p and 1080p videos

Qi Han Wong you’re absolutely correct. That’s why we built a five model architecture. We introduced a classification layer first. https://mindhyve.ai/architecture/