4 ms·

> I don’t see any path from continuous improvements to the (admittedly impressive) ‘machine learning’ field that leads to a general AI any more than I can see a

> I don’t see any path from continuous improvements to the (admittedly impressive) ‘machine learning’ field that leads to a general AI any more than I can see a path from continuous improvements in horse-breeding that leads to an internal combustion engine. While I also don't expect that AGI will emerge solely through optimizing statistical inference models, I also don't think "improvements to the machine learning field" consist only of such optimizations. Surely further insights, paradigm shifts, etc., will continue to play a role in advancing AI. Perhaps it's more a matter of semantics and a bad analogy; "machine learning" seems far more broad a field than "horse-breeding." Horse-breeding is necessarily limited to horses. Machine learning is not limited to a specific algorithm or data model. Even calling it a "statistical inference tool", while not wrong, is deceptive. What exactly does he or anyone expect or want an AGI to do that can't be understood at some level as "statistical inference"? One might say: "Well, I want it to actually understand or actually be conscious." Why? How would you ever know anyway?

mirekrusin 5y ago
It gets philosophical quickly, is "consciousness" repeatedly modifying cloud of random floats?
Jetrel 5y ago
It's worth pointing out that "machine learning" is a specific term of art, not a term for AI in general. It refers very specifically to the type of "convolutional neural networks" that have made a bunch of progress over the past 15-25 years. The moment you have a paradigm shift, sure, it can be considered "learning done by machines", but it's not "Machine Learning™" anymore. -- This is why the author put it in quotes; because, since it's a term comprehensible to anybody, it's got this unfortunate side effect where people on the outside of the field take the "plain english" meaning of it rather than realizing it's loaded with some extra specific meaning for the practitioners in the field.
- shannifin 5y ago
  > It's worth pointing out that "machine learning" is a specific term of art, not a term for AI in general. It refers very specifically to the type of "convolutional neural networks" that have made a bunch of progress over the past 15-25 years. In my experience, "machine learning" is more broad than "convolutional neural networks", aligning with Wikipedia's definition: "the study of computer algorithms that improve automatically through experience and by the use of data." https://en.wikipedia.org/wiki/Machine_learning https://en.wikipedia.org/wiki/Machine_learning

mirekrusin 5y ago

It gets philosophical quickly, is "consciousness" repeatedly modifying cloud of random floats?

Jetrel 5y ago

It's worth pointing out that "machine learning" is a specific term of art, not a term for AI in general. It refers very specifically to the type of "convolutional neural networks" that have made a bunch of progress over the past 15-25 years. The moment you have a paradigm shift, sure, it can be considered "learning done by machines", but it's not "Machine Learning™" anymore. -- This is why the author put it in quotes; because, since it's a term comprehensible to anybody, it's got this unfortunate side effect where people on the outside of the field take the "plain english" meaning of it rather than realizing it's loaded with some extra specific meaning for the practitioners in the field.

shannifin 5y ago
> It's worth pointing out that "machine learning" is a specific term of art, not a term for AI in general. It refers very specifically to the type of "convolutional neural networks" that have made a bunch of progress over the past 15-25 years. In my experience, "machine learning" is more broad than "convolutional neural networks", aligning with Wikipedia's definition: "the study of computer algorithms that improve automatically through experience and by the use of data." https://en.wikipedia.org/wiki/Machine_learning https://en.wikipedia.org/wiki/Machine_learning