4 ms·

So I think that the delineation between the scientist working with the content, and the Engineers who actually provide the mechanics for it is very fair. If th

by jariel 6y ago

So I think that the delineation between the scientist working with the content, and the Engineers who actually provide the mechanics for it is very fair. If there is a question mark here - it's really how much value are we deriving from all of these data people? Where is all the ML that's changing our lives? Search, Alexa and TikTok, I can see it. In the future obviously vision systems for autonomous cars etc.. But I'm really wary about the heavily decreasing marginal returns after that. It will surely change the world, but I think in specific areas. Most of the entire field seems like an optimization on something rather than anything new. Washing Machines feed up immense amount of labour and toil. Alexa telling me the weather is not.

avs733 6y ago
Most engineering and science jobs aren't a binary as much as they are a spectrum. If the article is trying to make a point about skill development and diversification, I'm totally on board. Bifurcating the roles instead is going to be less effective. To the value point...my sense has been we are seeing the Webcommerce 1.0 bubble Machine Learning edition. Lots of uses of it, not all of them have value. I am excited for where we will be in 10 or 15 years, but I suspect the difference will be huge. If you put me to a guess, I would say better data handling practices and ethics will likely be the linchpins of value creation vs. using tools for the sake of tools.
serjester 6y ago
I used to work at a legacy automaker and you’d be shocked at how much ML has changed certain areas of the business. It used to take an entire department to sort warranty claims and it’s now mostly automated. Aluminum part defects are now spotted automatically on the plant floor. Don’t even get me started with telematics data. Most software isn’t consumer facing but just because you don’t see it doesn’t mean it’s not changing things around you. ML tends to be overhyped but your assessment is too pessimistic.
- mateo411 6y ago
  I wouldn't think of a system that would automate the processing of warranty claims as ML. That's mostly applying the policy/rules to each claim. However, finding defects in aluminum parts that involves using computer vision, would absolutely be a ML solution.
  - serjester 6y ago
    There's millions of claims and thousands of car parts with all sorts of underlying issues. Unfortunately a rule based approach isn't feasible.
bart_spoon 6y ago
The vast majority of applications of machine learning that is changing the world isn't happening on a consumer level. Its happening in factories, warehouses, farms, logistics chains, etc.

by jariel 6y ago

avs733 6y ago

Most engineering and science jobs aren't a binary as much as they are a spectrum. If the article is trying to make a point about skill development and diversification, I'm totally on board. Bifurcating the roles instead is going to be less effective. To the value point...my sense has been we are seeing the Webcommerce 1.0 bubble Machine Learning edition. Lots of uses of it, not all of them have value. I am excited for where we will be in 10 or 15 years, but I suspect the difference will be huge. If you put me to a guess, I would say better data handling practices and ethics will likely be the linchpins of value creation vs. using tools for the sake of tools.

serjester 6y ago

I used to work at a legacy automaker and you’d be shocked at how much ML has changed certain areas of the business. It used to take an entire department to sort warranty claims and it’s now mostly automated. Aluminum part defects are now spotted automatically on the plant floor. Don’t even get me started with telematics data. Most software isn’t consumer facing but just because you don’t see it doesn’t mean it’s not changing things around you. ML tends to be overhyped but your assessment is too pessimistic.

mateo411 6y ago
I wouldn't think of a system that would automate the processing of warranty claims as ML. That's mostly applying the policy/rules to each claim. However, finding defects in aluminum parts that involves using computer vision, would absolutely be a ML solution.
- serjester 6y ago
  There's millions of claims and thousands of car parts with all sorts of underlying issues. Unfortunately a rule based approach isn't feasible.

bart_spoon 6y ago

The vast majority of applications of machine learning that is changing the world isn't happening on a consumer level. Its happening in factories, warehouses, farms, logistics chains, etc.