[olalonde]

While I agree that ANNs are probably a poor model for how the brain really works, your comparison isn't necessarily a fair one. While ANNs start their learning with an empty/random state, the human brain starts with a bunch of connections that are pre-encoded in DNA.

With that said, my guess is that we'll probably have to take a few more hints from the biology of the brain before we are able to achieve "human level intelligence".

[hansword]

You seem to have completely missed that NNs run on physical hardware.

A contemporary CPU and/or GPU is 'a bunch of connections that are pre-encoded' - just not in DNA, but in silicon.

[olalonde]

> A contemporary CPU and/or GPU is 'a bunch of connections that are pre-encoded' - just not in DNA, but in silicon.

No, CPUs/GPUs are not relevant if we are interested in the speed of learning in relation to the quantity of stimuli processed. You could even compute an ANN's learning algorithm with a pen and paper and it wouldn't change its "learning speed" within that definition. Pre-trained weights would.

Regardless, this head start is probably not sufficient to explain the disparity between the brain's capacity to learn and modern ANNs. ANNs are probably just not a very good approximation of how the brain works, for now at least.

[hansword]

I read this four times now, but I don't understand what you are saying.

[olalonde]

Can you be more specific? What I meant is this: if we postulate that the brain really is just like an ANN, it would be more like an ANN with pre-trained weights, thanks to evolution. In contrast, an ANN is typically initialized with random weights. A pre-trained network learns a lot faster than a randomly initialized one.