Browse Comments — Clean (de-noised)

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.

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.

Revise your curriculum, every university must inlcude AI courses like its a mandatory mathematics. It should be done through government ministries to re,evaluate all fields of studies. Otherwise there will be less humans ready for these new 1.7 M jobs.

Great point, Paul. The shift from model capability to workflow integration is a natural evolution. Minimizing tool switching allows teams to focus entirely on execution rather than managing disjointed subscriptions.

I think this is spot on, Paul Storm! For me, saying “RIP ChatGPT” isn’t about the tool dying, it’s about finally moving beyond one-off prompting to real systems and better orchestration. As I say in my keynotes, the breakthroghs come when we stop using AI like a fancy search box and start building with it

There is something practical about platforms trying to reduce friction instead of only competing on features or model benchmarks. People usually work better when ideas can move more smoothly from thought to execution.

Shift away from standalone tools toward unified environments reflects a broader pattern in software history. Integration usually solves friction on surface level while redistributing complexity underneath. Users often gain speed, yet lose some visibility into how outputs are assembled across models.

Value in modern AI ecosystems is increasingly tied to end-to-end workflow design, Paul. Efficiency gains tend to emerge when multiple capabilities operate within a single environment.

I wonder if the outcome would have been different if the prompt had specified that the patient was an expat living in the U.S. To me, the model’s behavior seems fairly logical. If a symptom description is written in Japanese, Chinese, or Hindi and no location is provided, the most likely assumption is that the patient is located in a region where that language is commonly spoken. Healthcare systems, care pathways, and thresholds for recommending the ER vary significantly across countries. This becomes even more interesting with languages that are spoken across many regions. French may point to France, Belgium, Switzerland, Quebec, or several African countries. Spanish could mean Mexico, Argentina, Spain, Colombia, or many others. Even English spans countries with very different healthcare practices. The real question may not be whether the model is culturally profiling the user, but whether it should be making geographic assumptions at all. In cases where location materially affects the recommendation, asking for location first might be the safer approach.

People rarely leave tools they outgrow, they leave the friction between tabs, prompts, and unfinished work.

Yeah I have to build a compute mesh connecting all of my neighbors and their compute together. If I connected all gaming computers in the world together AND invented an architecture to run efficiently on it that’s about 100x the worlds top 100 datacententers. Then you could simulate a human brain and shit. And gaming like this would be trivial - just need to invent the infrastructure and get people to trust eachother enough to integrate the economy of the system...

The consolidation angle is what makes this interesting, not just the features. Paul Storm

Removing friction is the name of the game in AI. It's all about making things faster and easier.

Every major technology shift follows the same pattern. People first dismiss it, then mock it, then quietly realize the people using it effectively are moving faster than everyone else.