Great info and summary; thank you, Bryan. I just did a video on AI supposedly running out of data and slowing down: https://youtu.be/exm56cPVuTc I argue that it is untrue and more of a schematics issue. I really liked that you brought up "microAI" in that this will be important for all mobile apps so that we will have AI at all times, even without the Internet. Once we have AI robots everywhere, the major issue will be human relationships with the AI. We will need to seriously address aspects of appropriate relationships and societal norms.
How AI agents interact is a huge unopen question. Error handling alone is mind-boggling to consider. Deploying these things in real world applications will be… interesting.
Much goodness to parse here, I gave Mistral a try for image descriptions, have to admit being impressed. Shared my examples, especially in comparison to my own hand written ones.
One thing I have not dug enough into- Mistral ius billed as open, indeed it releases its model weights (what would a normal person ever do with them?), but the thing most people seem to want to know is, what data is it trained on? Have I not scraped deep enough into Hugging Face?
Not only comprehensive, but well written; building from simple intro and examples of common agents to crucial considerations of sociocultural and technological factors fast coming into play. Galvanizing. Interesting to see how your integral experience and thought has been developing season to season, Bryan...
Great info and summary; thank you, Bryan. I just did a video on AI supposedly running out of data and slowing down: https://youtu.be/exm56cPVuTc I argue that it is untrue and more of a schematics issue. I really liked that you brought up "microAI" in that this will be important for all mobile apps so that we will have AI at all times, even without the Internet. Once we have AI robots everywhere, the major issue will be human relationships with the AI. We will need to seriously address aspects of appropriate relationships and societal norms.
I look forward to your valuable videos, as ever!
Interesting point about AI robots. I wonder which form factor will work.
Very comprehensive summary. Thanks for sharing.
How AI agents interact is a huge unopen question. Error handling alone is mind-boggling to consider. Deploying these things in real world applications will be… interesting.
I hope we do it better than IoT.
Much goodness to parse here, I gave Mistral a try for image descriptions, have to admit being impressed. Shared my examples, especially in comparison to my own hand written ones.
https://connect.oeglobal.org/t/impressive-image-descriptions-from-mistral-ai/7182
One thing I have not dug enough into- Mistral ius billed as open, indeed it releases its model weights (what would a normal person ever do with them?), but the thing most people seem to want to know is, what data is it trained on? Have I not scraped deep enough into Hugging Face?
You are doing the work of 5 bears!
Not only comprehensive, but well written; building from simple intro and examples of common agents to crucial considerations of sociocultural and technological factors fast coming into play. Galvanizing. Interesting to see how your integral experience and thought has been developing season to season, Bryan...