>Supposedly the next big thing in tech
>Everyone is jizzing over AI generated everything and predictions
>Nobody seems to be learning how to make an AI model
Why? Every coder/tech guy i ask about something like Tensorflow or Pytorch is like
>what
Why not learn it? Seems like easy cash after the you pass the initial learning curve as everyone wants to get into AI
Tip Your Landlord Shirt $21.68 |
DMT Has Friends For Me Shirt $21.68 |
Tip Your Landlord Shirt $21.68 |
I'm waiting for the AI IDE to do it for me.
if you dont know liner algebra dont bother
You aren't actually learning AI if you're using some framework
If I'm ever going to learn to do AI, I want to know how it actually works
Cnile thinking reinventing the wheel but badly is "how it works".
I know exactly how a hammer works without having build one. why can't you?
It's not really that complicated of a concept. Think of a simple feed forward neural net as being a psuedo-random number generator with a bunch of internal parameters that can be tweaked until the outputs for a set of inputs is useful. That's basically it. Now go ham.
>if statements = AI
This trend toward fricking "AI" bullshit when it's just simple goddamn machine algorithms is so goddamn cringe and aimed toward fricking computer-phobic idiots. It's horrible.
Technically the human brain is also an if machine so you know...
It is also questionable whether people are generally intelligent. Trying to claim a machine is, is just that much more out there.
Not really. It uses impressions of things we sense and a very fuzzy sort of pattern matching that cannot be emulated in machines to create intelligence. It's not just a bunch of if/then blocks that use a known algorithm to do something people erroneously attribute to "intelligence", artificial or no.
>that cannot be emulated in machines to create intelligence
Souls don't exist you fricking moron
>humans=my python app
Maximum darling krugr.
ai was perfected with chess. everything after is downgrade
Because getting quality data sets is a pain in the ass and I can't be arsed.
Most all of it is marketing. It's expensive toget valuable datasets, processing them is very expensive, and the results are very lackluster.
No wonder people actually working with computers aren't interested.
>Most all of it is marketing
Lmao cope, give it a year and your post on 4chin will go through its AI NN to check if everything is "ok" with it just like its done on facebook right now
If you've lived longer than google, you'll know different things throughout history have been marketed as "AI". It never is. The current statistical tools clever engineers use to solve specific problems without specific handcoding is cool, but is a sorry excuse for an AI. There is nothing intelligent about it.
I am not a fan of calling it AI but it is pretty fricking good for generating things, years ago you couldnt do shit like DALL can now with its "Neural Network" like structure.
Generating statistically likely images or text isn't something new we've just discovered. We've got the horsepower to do what we've already known for a long time. It's useful in a lot of specific scenarios, but they're just so far from what it's marketed to be.
Your comment about facebook is really sabotaging your point. Favebook's checks can't differentiate between porn and onions, even with tools worth more than 100M. They still use huge swaths of moderators to pull the strings.
>We've got the horsepower to do what we've already known for a long time
Well yeah the whole Neural Network shit was invented decades ago but we finally can use it so its relevant now.
people are learning it, how else do you thnk its made moron