3 ms·

I shifted to doing AI (mostly vision) stuff a few years back, and I sometimes feel I was happier doing backend systems. There is so much noise out there in AI a

I shifted to doing AI (mostly vision) stuff a few years back, and I sometimes feel I was happier doing backend systems. There is so much noise out there in AI and the whole thing is moving at the pace of JS frameworks (i.e. too fast to humanly stay up-to-date with). I don't know if it is generally better to make the switch to AI or stick with the old stuff. People will always need the old stuff. Also, there is a reason those AI projects have gotten canceled at your workplace btw .. I think most orgs don't have the right setup to derive value from AI. Question to OP if you want to discuss this .. how do you see ML/DS vs. vision. Vision needs a ton of data and I think hard for a lot of orgs to get benefit from. Is ML/DS very different?

rafiki6 6y ago
Agreed on most of what you said. DS doesn't necessarily need a lot of data. Alot of DS work is not much more than visualizations and report generation to find insights. ML might need a lot of data depending on how your org defines it. There are many interesting non-deep learning applications out there. At my org I was mostly involved in NLP work. Honestly I think that will be the big game changer for most of the Enterprise, as they primarily deal with text data. Think automated parsing and analysis of emails, chats, reports etc. Something like GPT-3, can be a game changer if/when it doesn't need a multi billion dollar super computer.