@filippie509@techhub.social
@filippie509@techhub.social avatar

filippie509

@filippie509@techhub.social

A pinch of skepticism against the AI hype machine since 2015. Enjoy the content after Musk musks twitter.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

filippie509, to random
@filippie509@techhub.social avatar

The main purpose to study history IMHO, is to see what kind of things millions of people, sometimes entire nations, believed unquestionably, which later turned out to be complete bonkers bullshit.

And then realize there is no reason whatsoever to assume today is any different.

filippie509, to ai
@filippie509@techhub.social avatar

Since a diagram is worth a 1000 words, and public is confused about what developments in mean for robotics, I took some time to draw this. Feedback welcomed.

filippie509, to ai
@filippie509@techhub.social avatar

What could possibly go wrong?

image/png

filippie509, to random
@filippie509@techhub.social avatar

It has been 3 years since a million Tesla robotaxis were 100% going to hit the road by the end of 2020. Boy what a ride that was indeed. Looking forward to the Mars city and VTOL electric jets. Oh and hyperloop of course.

filippie509, to random
@filippie509@techhub.social avatar

Been getting lot of these after publishing my latest post (here if you missed it: https://blog.piekniewski.info/2023/10/21/the-church-of-agi/ ). Turns out it's worse than I thought.

image/jpeg

filippie509, to ai
@filippie509@techhub.social avatar

Every freaking time...

image/png

filippie509, to random
@filippie509@techhub.social avatar

This Sunday is election in Poland. I wouldn't bother my US followers with this, but this time, the ruling party is pulling such unprecedented shenanigans to stay in power, that this election will not be about politics and more about whether Poland remains a liberal democracy.

filippie509, to random
@filippie509@techhub.social avatar

Look!!!! A Miracle! A Miracle!...

Except...

There are literally thousands upon thousands of websites that put this exact image in the context of plot of Inception, including alt text. This AI craze is getting really stupid.

filippie509, to random
@filippie509@techhub.social avatar

Is generative AI fun to play with? Yes.
Is it amusing and addicting? Yes.
Can it be useful in some cases? Yes.
Is it the revolution at the scale of electricity or Industrial Revolution? F**** NO!

filippie509, to random
@filippie509@techhub.social avatar

This is a fascinating culture clash: break things and fail fast, sloppy SV approach VS meticulously prove your sw formally. Both have a place in the world IMHO, but it's fascinating how SV guys cannot comprehend that SW can actually be provably bug-less.

filippie509, to random
@filippie509@techhub.social avatar

Machines are precise, deterministic and repeatable. But can't tolerate randomness on input. Humans are sloppy and non-deterministic but robust to wide spectrum of randomness. ANNs can tolerate slightly more randomness but aren't very precise, hardly repeatable and not very robust. Which makes ANNs suitable for non-critical applications in which we want to automate something noisy, but we can either well characterize the "spread" of input randomness or don't need to worry when model fails. So we can use these for e.g. ad placement, artistic tools, internet search, some limited perception tasks in industrial environments and so on. There are actually plenty applications within this cognitive "middle ground". But be careful to try to replace a human in the context of broad randomness. E.g. driving. Anything manual. Anything where cost of failure is high. Anything with potential liability for mistakes such as medical applications.

filippie509,
@filippie509@techhub.social avatar

Those are exactly the applications that sci-fi writers, visionaries and some VC's are most excited about. But from a purely technical point of view, unfortunately we are at the stage of Otto Lilienthal rather than Orville Wright, and so anybody trying that will break their neck. That is not say we should not try - but we have to be asking the right questions. That was the foundation of success of Wright brothers. They finally understood the basics of aerodynamics and hence managed to build a flying machine. But they were organized and methodical.

filippie509, to random
@filippie509@techhub.social avatar

The elephant in the room of AI wonders: https://www.honest-broker.com/p/ugly-numbers-from-microsoft-and-chatgpt

filippie509, to random
@filippie509@techhub.social avatar
filippie509, to random
@filippie509@techhub.social avatar

Is this rock bottom yet as far as dumb stuff from SV?

filippie509, to random
@filippie509@techhub.social avatar

Mixing up the current boom in machine learning with the birth of actual Artificial Intelligence is exactly like mixing up the boom in alchemy with the birth of nuclear physics. The promises are similar & some of the motions are the same, but it probably needs a few centuries.

filippie509, to ai
@filippie509@techhub.social avatar

For all those who think all those influencers are genuinely excited about GPT etc, here is how that sausage is made:

filippie509, to random
@filippie509@techhub.social avatar

LLMao effect - I keep hearing from influencers that LLMs are about take millions of jobs. But when I try to figure out which exactly jobs are ready to be displaced by LLMs I'm always left with influencers.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • โ€ข
  • megavids
  • kavyap
  • DreamBathrooms
  • thenastyranch
  • magazineikmin
  • InstantRegret
  • GTA5RPClips
  • Youngstown
  • everett
  • slotface
  • rosin
  • osvaldo12
  • mdbf
  • ngwrru68w68
  • JUstTest
  • cubers
  • modclub
  • normalnudes
  • tester
  • khanakhh
  • Durango
  • ethstaker
  • tacticalgear
  • Leos
  • provamag3
  • anitta
  • cisconetworking
  • lostlight
  • All magazines
    •