Why Apple believes it’s an AI leader–and why it says critics have it all wrong

Post by SilasX » Sat, Aug 08, 2020, 5:37 PM UTC

parekhnish wrote: post #189

jointpdf wrote: post #34

Setting aside all the other work Apple is doing in this area, can someone explain why iMessage autocorrect is unacceptably bad (to put it diplomatically) in the era of weapons-grade language models? Its behavior is totally bizarre. It’s like an underpowered chess engine that makes flagrant blunders: capitalizing random words in the middle of a sentence (rock -> Rock), the context sensitivity of an actual rock, forcin…

> ... in the era of weapons-grade language models? Just an aside: Although common perception is that "weapons-grade" refers to the highest quality, in reality it refers to the lowest-possible quality that meets the bare minimum requirements!

Post by m463 » Sat, Aug 08, 2020, 6:01 PM UTC

tomgp wrote: post #71

jointpdf wrote: post #34

Setting aside all the other work Apple is doing in this area, can someone explain why iMessage autocorrect is unacceptably bad (to put it diplomatically) in the era of weapons-grade language models? Its behavior is totally bizarre. It’s like an underpowered chess engine that makes flagrant blunders: capitalizing random words in the middle of a sentence (rock -> Rock), the context sensitivity of an actual rock, forcin…

I once mentioned the rapper xxxtentacion in a text message. Now whenever I sign of an email or text to my wife with an ‘x’ iOS decides that what I actually meant to do was write ‘xxxtentacion’ and I’ve normally pressed send before I notice. After a couple of years now this is just how I talk to my wife.

Post by m463 » Sat, Aug 08, 2020, 6:06 PM UTC

lacker wrote: post #84

"Google is an amazing company, and there's some really great technologists working there," he said. "But fundamentally, their business model is different and they're not known for shipping consumer experiences that are used by hundreds of millions of people." What does he mean by this? Google search and Android are both used by more than a billion people. YouTube over 2 billion. These are all bigger than any Apple pr…

Post by antipaul » Sat, Aug 08, 2020, 6:33 PM UTC

moandcompany wrote: post #23

Despegar wrote: post #3

This piece is interesting because Apple was saying this all along but no one really believed them because it sounded like excuse making. But here JG is basically saying the same thing: >Yes, I understand this perception of bigger models in data centers somehow are more accurate, but it's actually wrong. It's actually technically wrong. It's better to run the model close to the data, rather than moving the data around…

When it comes to machine learning, the subtlety here is that there are at least two sides or facets to machine learning: (1) training and (2) inference. It's fair to say that there are multiple areas for AI leadership. It is generally believed that: Model Creation (1) Those with the access to the best (which is not necessarily the most, but often believed to be) data have a strong starting point for training models;…

Post by hydroreadsstuff » Sat, Aug 08, 2020, 6:43 PM UTC

jointpdf wrote: post #34

Setting aside all the other work Apple is doing in this area, can someone explain why iMessage autocorrect is unacceptably bad (to put it diplomatically) in the era of weapons-grade language models? Its behavior is totally bizarre. It’s like an underpowered chess engine that makes flagrant blunders: capitalizing random words in the middle of a sentence (rock -> Rock), the context sensitivity of an actual rock, forcin…