5 ms·

Agreed. To my mind, this is one of the things that makes it so hard to REALLY achieve artificial intelligence - we don't even know exactly what we're supposed t

Agreed. To my mind, this is one of the things that makes it so hard to REALLY achieve artificial intelligence - we don't even know exactly what we're supposed to be creating. It seems to me that CS types who are actually working on AI (as opposed to sophisticated machine learning) are going to need some help from philosophers to even fully define the problem.

dchichkov 14y ago
As 'CS type' actually working on ML, I disagree. I'd rather prefer some help from hardware guys, that would allow me to run a neuro network with 100 billion neurons really really fast.
- dmix 14y ago
  If it's your long term dream of fully replicating humans, it would be helpful to read Kant and similar philosophers to have a clear understanding of what consciousness is. If it's not and the goal is just replicating and accelerating certain cognitive abilities, then it might not be super helpful and possibly disheartening (disheartening because of how much we lack full understanding of consciousness due to limitations of only being able to understand it via our own consciousness).
  - rosser 14y ago
    My undergraduate degree was in philosophy, with a concentration in philosophy of mind, and while I can't recommend highly enough taking the time to read through and learn from the sundry perspectives and takes we've had on consciousness throughout history, be warned that that way lies confusion and dragons. Kant certainly didn't know what consciousness was. He had some instructive things to say, and massively helped frame the debate for people who came later, but even present-day philosophers who have the benefit of modern science and technology to help in their investigation are no closer to pinning down how it is that meat can feel.
  - aba_sababa 14y ago
    Absolutely not. After 2500 years, Western philosophy has yet to achieve broad consensus on anything at all. Philosophers don't have a clue what consciousness is - they can barely even describe it without using utterly ambiguous words like "qualia" or "sense datum". The only way we'll ever be able to know anything about consciousness besides what's going on inside our own heads is for us to directly experience what's going on inside someone else's (or some animal's). Not until the Singularity, basically. I'm not going to hold my breath for that one.
  - lutusp 14y ago
    > If it's your long term dream of fully replicating humans, it would be helpful to read Kant and similar philosophers to have a clear understanding of what consciousness is. From a scientific perspective, this elevates philosophy beyond anything reasonable. Philosophers have many things to say, but they tend not to be the kind of people who are trained to ask, "Is there an empirical, falsifiable test of my claim?"
- fusiongyro 14y ago
  Throw all the hardware you want at it. Without philosophy you have no way to answer the question: what does it mean? Do you think you can just simulate 100 billion neurons and then ask the computer?
  - pygy_ 14y ago
    > what does it mean? Assuming that it means something to begin with...
  - Locke1689 14y ago
    The question of whether a computer can think is no more interesting than whether a submarine can swim Call it what you want, if it works, that's all that matters. Also, presuming that a "thinking" brain must be like a human brain is naive.

dchichkov 14y ago

As 'CS type' actually working on ML, I disagree. I'd rather prefer some help from hardware guys, that would allow me to run a neuro network with 100 billion neurons really really fast.

dmix 14y ago
If it's your long term dream of fully replicating humans, it would be helpful to read Kant and similar philosophers to have a clear understanding of what consciousness is. If it's not and the goal is just replicating and accelerating certain cognitive abilities, then it might not be super helpful and possibly disheartening (disheartening because of how much we lack full understanding of consciousness due to limitations of only being able to understand it via our own consciousness).
- rosser 14y ago
  My undergraduate degree was in philosophy, with a concentration in philosophy of mind, and while I can't recommend highly enough taking the time to read through and learn from the sundry perspectives and takes we've had on consciousness throughout history, be warned that that way lies confusion and dragons. Kant certainly didn't know what consciousness was. He had some instructive things to say, and massively helped frame the debate for people who came later, but even present-day philosophers who have the benefit of modern science and technology to help in their investigation are no closer to pinning down how it is that meat can feel.
- aba_sababa 14y ago
  Absolutely not. After 2500 years, Western philosophy has yet to achieve broad consensus on anything at all. Philosophers don't have a clue what consciousness is - they can barely even describe it without using utterly ambiguous words like "qualia" or "sense datum". The only way we'll ever be able to know anything about consciousness besides what's going on inside our own heads is for us to directly experience what's going on inside someone else's (or some animal's). Not until the Singularity, basically. I'm not going to hold my breath for that one.
- lutusp 14y ago
  > If it's your long term dream of fully replicating humans, it would be helpful to read Kant and similar philosophers to have a clear understanding of what consciousness is. From a scientific perspective, this elevates philosophy beyond anything reasonable. Philosophers have many things to say, but they tend not to be the kind of people who are trained to ask, "Is there an empirical, falsifiable test of my claim?"
fusiongyro 14y ago
Throw all the hardware you want at it. Without philosophy you have no way to answer the question: what does it mean? Do you think you can just simulate 100 billion neurons and then ask the computer?
- pygy_ 14y ago
  > what does it mean? Assuming that it means something to begin with...
- Locke1689 14y ago
  The question of whether a computer can think is no more interesting than whether a submarine can swim Call it what you want, if it works, that's all that matters. Also, presuming that a "thinking" brain must be like a human brain is naive.