3 ms·

i think the back reaction against AI/ML has gone too far if this phraseology becomes the locus of your criticism. > Sure, there are likely some traditional ML

i think the back reaction against AI/ML has gone too far if this phraseology becomes the locus of your criticism. > Sure, there are likely some traditional ML models and techniques involved in this work. It first launched in 2021 so it's unlikely there were any generative models (which, let's face it, is what many people in 2024 assume when they see "AI" mentioned). I don’t think you have a correct mental model of the history of the field. “Generative” is not a very clear technical term and most ML from this era is not going to be “traditional”/classical - it’s almost certainly still DL

simonw 2y ago
Does the term "traditional" not imply deep learning? Genuine question, I wasn't aware that "traditional" had a specific meaning - I'm using it here to reference pre-current-spike-in-generative.
- whimsicalism 2y ago
  I don’t think the distinction between “generative” vs deep is a super defensible/technical one - compared to the normal distinction made between classical & deep - where deep is about high parameter multilayer networks (around since 2016) and classical refers to linear models and tree based models preceding them. There is a classic distinction between generative and discriminative models but it does not fully map on to what people mean when they say “generative” nowadays
- sib 2y ago
  The study / field of AI goes back decades further than deep learning, so, no, I would say that "traditional" AI does not imply deep learning.
  - simonw 2y ago
    If we are going by the original definition of AI from 1956 that's "make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves". I don't think that applies to this Google project: no human could perform the work needed to optimize traffic lights to reduce stop/starting, the dataset they would have to consider is way too large.
  - seadan83 2y ago
    Agree, to add, there are many different AIs that have been created over time. The "classical" AI is not a monolith and "classical" vs LLM or ML is seemingly overly reductive. For example, decision tree AI vs weighted heuristic, etc.. There are many classes of AI beyond ML and LLM

simonw 2y ago

Does the term "traditional" not imply deep learning? Genuine question, I wasn't aware that "traditional" had a specific meaning - I'm using it here to reference pre-current-spike-in-generative.

whimsicalism 2y ago
I don’t think the distinction between “generative” vs deep is a super defensible/technical one - compared to the normal distinction made between classical & deep - where deep is about high parameter multilayer networks (around since 2016) and classical refers to linear models and tree based models preceding them. There is a classic distinction between generative and discriminative models but it does not fully map on to what people mean when they say “generative” nowadays
sib 2y ago
The study / field of AI goes back decades further than deep learning, so, no, I would say that "traditional" AI does not imply deep learning.
- simonw 2y ago
  If we are going by the original definition of AI from 1956 that's "make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves". I don't think that applies to this Google project: no human could perform the work needed to optimize traffic lights to reduce stop/starting, the dataset they would have to consider is way too large.
- seadan83 2y ago
  Agree, to add, there are many different AIs that have been created over time. The "classical" AI is not a monolith and "classical" vs LLM or ML is seemingly overly reductive. For example, decision tree AI vs weighted heuristic, etc.. There are many classes of AI beyond ML and LLM