Any tips or tools to share? Help us all 10x our productivity.
6
I'm not. If you're trying to wedge AI/ML in to a workflow and you can't think how, you don't need it.
Ignore fads.
2
I don't think AI is at a level that it would just generally improve something once applied. But it's awesome for a select few tasks.
I saw that not many people used AI for text analysis, so I built https://wiseer.io
It's a text analysis tool, and not only because I made it but I really think that if you have feedback/comments or similar data you should run it through some analysis tool so you don't have to read everything yourself. Or at the very least, to get an idea of the general level of feedback at a glance.
If the source of that data is continuously filled it gets even more important to have some kind of service look through it rather than yourself.
2
You don't have to force yourself on the bandwagon just because it's the absolute craze right now.
And project managers now talk about having to use "AI" when all it takes are a couple IF statements. Hey if that's AI for ya, great!
1
I use it trough services, other people build.
Like OCR, Voice transcription, etc.
This is most convenient. ;)
1
I often need to use a trained model in a different language than my go-to ML language (python).
Some examples I use:
scikit-learn lin-reg, you can obtain coefficients and re-write in other language.
scikit-learn decision tree: just look at the tree and re-write in other language.
keras: save weights and compute forward propagation in other language.
Often copying these weights doesn't cost too much time, but if there is a better way, I'm up for it :).
1
I do like your line of thought. 🙂
We are developing an assistant for both personal and business affairs which includes components enhanced with advanced automation and AI.
People and businesses actually generate so much metadata that can be used ML models to better understand what's going on.
1
Probably hard to find uses without a large amount of data to train from.
1
Most of the people talk about AI but reality is logistic regression :)
I'm not. If you're trying to wedge AI/ML in to a workflow and you can't think how, you don't need it.
Ignore fads.
I don't think AI is at a level that it would just generally improve something once applied. But it's awesome for a select few tasks.
I saw that not many people used AI for text analysis, so I built https://wiseer.io
It's a text analysis tool, and not only because I made it but I really think that if you have feedback/comments or similar data you should run it through some analysis tool so you don't have to read everything yourself. Or at the very least, to get an idea of the general level of feedback at a glance.
If the source of that data is continuously filled it gets even more important to have some kind of service look through it rather than yourself.
You don't have to force yourself on the bandwagon just because it's the absolute craze right now.
And project managers now talk about having to use "AI" when all it takes are a couple IF statements. Hey if that's AI for ya, great!
I use it trough services, other people build.
Like OCR, Voice transcription, etc.
This is most convenient. ;)
I often need to use a trained model in a different language than my go-to ML language (python).
Some examples I use:
scikit-learn lin-reg, you can obtain coefficients and re-write in other language.
scikit-learn decision tree: just look at the tree and re-write in other language.
keras: save weights and compute forward propagation in other language.
Often copying these weights doesn't cost too much time, but if there is a better way, I'm up for it :).
I do like your line of thought. 🙂
We are developing an assistant for both personal and business affairs which includes components enhanced with advanced automation and AI.
People and businesses actually generate so much metadata that can be used ML models to better understand what's going on.
Probably hard to find uses without a large amount of data to train from.
Most of the people talk about AI but reality is logistic regression :)
https://en.wikipedia.org/wiki/Logistic_regression