Great news everyone! I finally talk about AI hype. Someone finally mentioned LLMs one time too many, and the reckoning is upon us:
https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/
Great news everyone! I finally talk about AI hype. Someone finally mentioned LLMs one time too many, and the reckoning is upon us:
https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/
@ludicity refreshing article, takes me back to my corporate days leading digitalization strategy for supply chain operations. C suite always wanted a magical AI story. When I detailed the use case the team had unlocked in one of the regions, and also detailed the effort to reap the benefits, they were rather underwhelmed, despite the fact that the use case paid for itself in terms of the data engineers we needed to hire, and some other tangible benefits. The truth is great AI use cases are narrow, and require a lot of data engineering to get right. There's no magical shortcut that scales across the organization. I'm speaking for about 80% of companies, and I only called them AI because that's what the political jargon was, if you said algorithm or machine learning, most executives weren't interested in hearing it. And this was before LLMs showed up. I do wonder what the political landscape is like in a large organization nowadays
@noahjunior10 I'm genuinely slightly confused as to why more data scientists don't come with the entire data engineering brain bundled out of university, at least for systems dealing with <30GB data or whatever per month. It was like, REALLY easy to transition.
@ludicity it's the same mystery I could not figure out either, we weren't dealing with mega datasets. The teams that were close to the use cases were asking me for data engineering resources as the priority, saying the scientists weren't very productive as is. The broader organization was focusing recruiting strategy on data scientists, so we had to write new job descriptions for this category
@noahjunior10 There are only three dataset sizes:
1. Fits in memory
2. Fits in memory of several computers
3. Hire an Amazon senior