News for hackers. Stuff that matters.

A comment on Measuring abstract reasoning in neural networks.

Re:Measuring abstract reasoning in neural networks by tzahola on Sunday July 15, 2018 @07:35AM (#17534791)
I’m starting to get tired of this antropomorphized terminology for neural networks. “Belief propagation”, “abstract reasoning”... No wonder that laypeople think we’re on the brink of the AI apocalypse.
[ Parent | on HN ]
Re:Measuring abstract reasoning in neural networks by amelius on Sunday July 15, 2018 @08:05AM (#17534876)
Yes, sadly, even scientists use marketing-speak. And I guess companies use this style of terminology to attract VCs and to push their half-baked products.

Not sure if complaining about it would help though.

[ Parent | on HN ]
Re:Measuring abstract reasoning in neural networks by mindcrime on Monday July 16, 2018 @10:02AM (#17541286)
They are analogies, and there seems to be a lot of weight behind the idea that analogy making is core to human reasoning. So it would seem to make sense that we've adopted these analogies.

Besides, why bother creating a whole second, parallel, terminology?

[ Parent | on HN ]