This reminds me of the movie Leave no one behind. No one is actually in charge. No central authority. But it works. I think this is the direction many things should go including education and government. People are the cars in this example and so many things can map to the roadways.
I have a big question here. I'm working at a medical facility and the chief scientist is really dismissing my ideas of creating graphs because he thinks that chaining many instances of the same LLM are not able to generate an outcome better than the single llm just in term of quality. I disagree. Is there any hard evidence that graphs are more efficient in terms of outcome?
The topography described in this video, and interestingly the underlying control logic being explored, is very similar to the real world approach to national air traffic control systems, at least in the US. It’s quite uncanny.
As a commercial pilot working in the industry, I've noticed a huge untapped market in Air Traffic Control due to low tech adoption rates. Here's the opportunity I see: Air Traffic Control is highly procedural and scripted, with controllers following set protocols (HINT: CHAIN OF THOUGHT) most of the time. The goal isn't to replace human controllers - that would be unsafe. Instead, we should aim to augment their capabilities, dramatically increasing the amount of traffic they can manage at once. We have an incredible resource at our disposal: over 50 years of ATC recordings. With platforms like LiveATC making these recordings mainstream, we have access to an enormous amount of data for training AI systems. Ai generated radio calls on behalf of the ATC controller. Instead of speaking, he or she can queue radio calls. The controlled nature of airspace makes it incredibly predictable - 99.9925% of the time, in fact. This predictability lends itself well to AI assistance. We could implement an AI system to handle routine tasks and communications, while maintaining human agency for complex situations and exceptions. This approach has the potential to significantly improve both efficiency and safety in air traffic management. By leveraging AI for the routine aspects, we free up human controllers to focus on critical decision-making and handling unexpected events. If you read this far and are interested in exploring this let me know, maybe we need to start a community around it.
I can definitely see how this could be used to solve power transmission with renewable energy sources. Possibly the biggest obstacle in renewable adoption (in US at least) is transmission across old power grids. Rooftop solar and batteries are decentralized sources with two way flow. Of course renewable generation is not consistent - depends on when the wind blows and sun shines.
You can either present the formal theory of a new method by introducing all relevant parameters (see my presentation of all parameters) or you can choose to go another way and - like I did - focus to explain the core idea of new methods, without applying a pure mathematical formal description of the system. And I want to make these video as easy as possible to understand to new-comers to AI.
This reminds me of the movie Leave no one behind. No one is actually in charge. No central authority. But it works. I think this is the direction many things should go including education and government. People are the cars in this example and so many things can map to the roadways.
I have a big question here. I'm working at a medical facility and the chief scientist is really dismissing my ideas of creating graphs because he thinks that chaining many instances of the same LLM are not able to generate an outcome better than the single llm just in term of quality. I disagree. Is there any hard evidence that graphs are more efficient in terms of outcome?
I''ve never been the first to comment anywhere! keep up the great work!
Nice! 🙌🏾
The topography described in this video, and interestingly the underlying control logic being explored, is very similar to the real world approach to national air traffic control systems, at least in the US. It’s quite uncanny.
As a commercial pilot working in the industry, I've noticed a huge untapped market in Air Traffic Control due to low tech adoption rates. Here's the opportunity I see:
Air Traffic Control is highly procedural and scripted, with controllers following set protocols (HINT: CHAIN OF THOUGHT) most of the time. The goal isn't to replace human controllers - that would be unsafe. Instead, we should aim to augment their capabilities, dramatically increasing the amount of traffic they can manage at once.
We have an incredible resource at our disposal: over 50 years of ATC recordings. With platforms like LiveATC making these recordings mainstream, we have access to an enormous amount of data for training AI systems.
Ai generated radio calls on behalf of the ATC controller. Instead of speaking, he or she can queue radio calls.
The controlled nature of airspace makes it incredibly predictable - 99.9925% of the time, in fact. This predictability lends itself well to AI assistance. We could implement an AI system to handle routine tasks and communications, while maintaining human agency for complex situations and exceptions.
This approach has the potential to significantly improve both efficiency and safety in air traffic management. By leveraging AI for the routine aspects, we free up human controllers to focus on critical decision-making and handling unexpected events.
If you read this far and are interested in exploring this let me know, maybe we need to start a community around it.
@@Max-hj6nqthis is a brilliant idea!!
@@wdonno it got deleted :(
Just an fyi if you are like me and like to know: in English asynchronous is pronounced a sink CRO nis
Omg this is my favorite subject swarm ecosystems are what got me interested in AI 😍🧠
I can definitely see how this could be used to solve power transmission with renewable energy sources. Possibly the biggest obstacle in renewable adoption (in US at least) is transmission across old power grids.
Rooftop solar and batteries are decentralized sources with two way flow.
Of course renewable generation is not consistent - depends on when the wind blows and sun shines.
Not clear why you introduced the slide with all parameters from all 3 papers?
You can either present the formal theory of a new method by introducing all relevant parameters (see my presentation of all parameters) or you can choose to go another way and - like I did - focus to explain the core idea of new methods, without applying a pure mathematical formal description of the system. And I want to make these video as easy as possible to understand to new-comers to AI.
Thank you for the executive summary in the description. It's a fascinating subject!
Glad it was helpful!