Hacker News Reader

Comment by cellular | Hacker News Reader

It seems like at least half the tech people I talk to work with AI now... no matter their field.

Any type of heuristic search = AI. hand-made decision tree or lookup table = AI, Naive bayes = AI, KNN = AI. Uses numpy = AI. One employee in mobile app startup has graduate thesis in ML = AI startup. Approximate string matching in SQL query = AI. Not clear what the product is going to be = AI.

"Information is not knowledge. Knowledge is not wisdom. Wisdom is not truth. Truth is not beauty. Beauty is not blockchain. Blockchain is not AI. AI is THE BEST.”

In my space (cybersecurity) I’ve heard “more than 2 joins” (in an SQL context) is ML.

So it's turning into a meaningless buzzword.

I worked for a company that had a "big data machine learning" product. Yeah, it was mostly elasticsearch aggregations.

well, aggregations are just leveraging how the data fields are analyzed and labeled, so if those were based on ML/NLP techniques then it could be legit

> so if those were based on ML/NLP techniques then it could be legit

Occam's Razor is whispering that it probably wasn't legit.

i believe the only NLP/ML stuff we really had was at the level of positive / negative sentiment analysis for social media posts, which as far as i know is pretty basic and mostly just involves dictionary checks for +/- words. the elasticsearch aggregations & indexing was not ML/NLP based at all.

it turns out significant amounts of what I did in the past overlaps significantly with ML now: for example, hyperparameter exploration. I created a system years ago to do exploration by fitting polynomial surfaces, finding predicted maxima, and exploring those areas. Also, many of the algorithms underlying the area I did my PhD research in (molecular dynamics) are very similar to gradient descent.

I mean, if the software is smart enough to make sense of long sequences of 1's and 0's then it must be artificially intelligent to some extent. /s

Reply on news.ycombinator.com