[enord]

This is the high water mark, the summer solstice if you will, of the current wave.

Buckle up, Serious People are going to rediscover the fundamental Hard Problems and relocate the current Hot Topics into their appropriate ontologies.

[zitterbewegung]

I can't tell if this is satire about the current state of AI and its hype cycle or not.

[text_exch]

It is referencing the following xkcd: https://xkcd.com/1831/

[curuinor]

yeah, cousin comments are treating it as if it's serious, but the capital letters make it sound sarcastic

[sgt101]

These are the serious people, look at josh's Google scholar and read the intuitive reasoning work.

[nopinsight]

What is some evidence or reasoning that we are reaching certain hard limits as you mentioned?

[jjaredsimpson]

I'll take a crack. Progress in ML applications to previously intractable problems has created an irrational optimism that AGI is on the near term horizon.

I'm not aware of any novel abstraction which led to a solution to these intractable problems. The problems became tractable because of silicon and incremental algorithm improvements.

This is another way of saying, yeah intractable problems are being made tractable, but these problems aren't stepping stones to AGI.

[randcraw]

For example, machine translation from one human language to another has been acclaimed as one of the big success areas in deep learning. But when one looks deeper, there be dragons...

https://www.theatlantic.com/technology/archive/2018/01/the-s...

[YeGoblynQueenne]

Oh wow. That's Douglas Hofstadter in great form. Could you please submit that to HN so it has a chance to get to the fist page?

[YeGoblynQueenne]

Never mind- there's already a conversation. Thanks for posting anyway.

[enord]

I never mentioned limits, and the reasoning is purely inductive, based on previous events.

We are far away from understanding intelligence in all directions. Top down, bottom up, neurologically, psychologically, logically, mathematically and last but not least philosophically.

There's optimism at the moment, because we are doing more stuff with more annotated data (the annotations providing the semantic grounding, as in "Not hot dog" vs. "Not in category 339492-883764-399274"). The key difference this time being access to (and processing power for-) sufficiently large "training sets" (read "samples") for deep-learning algorithms (read "statistical models"). From an AGI point of view, this is nothing but an expensive parlor trick, because the "intelligent" part is the annotation, not the categorization after the fact.

[YeGoblynQueenne]

The annotation is not even particularly smart. In supervised classification, labels are basically scalars, standing for... whatever the researcher means them to stand for. The class represented by the label can be as broad or as narrow as the researcher wants it to be. Even the relation of the class with the data it is supposed to represent is arbitrary and its choice entirely unprincipled and based on instinct alone.

Which goes to show that a) our machine learning models are dumb as bricks and b) they are as far from AGI as worms are from building a rocket to go to the moon, where their god lives (see all those holes up there?).

[ddorian43]

Please explain in simpler terms ?

[enord]

Winter is coming.

[justaman]

I'll take a stab at it.

I think they are saying...Intellectuals are going to change the way they guide their research or formulate their hypothesis based on the this work at MIT. This person believes that this project signifies a paradigm shift in the fundamental origins of thought, hereafter nothing will be the same.

I think this person is trying way to hard to sound smart or poetic.

[lucozade]

Nah. enord's saying that they'll all sit down together and realise that they don't know very much about intelligence after all.

So all the current excitement around super-intelligent AIs, and whatnot, will go the same way as it has the previous times we all got excited.

[logfromblammo]

Some of the most intelligent people in the world will gather together, and learn that they still don't even know what "intelligent" means, much less how to actually build it.

Then they will discover that this has happened before--several times--and the reasons they had to think that it was all different this time were all based on exaggeration and wishful thinking, promoted by people trying to make a buck.

We can throw more hardware at the problem now, but even so, every advance seems to be accompanied by a reassessment and lengthening of the distance to the finish line.

[k__]

But isn't AI getting percievably better every run?

[sampo]

The finish line of a marathon is getting nearer every step, but we've yet to clear even the first mile.

[fjsolwmv]

"this person" was channeling the pretentious buzzword-packed hype of the OP article.