World & Business

Home
Technology
Gaming
Entertainment
World & Business
Science
Sports
AI

Home
Technology
Gaming
Entertainment
World & Business
Science
Sports
AI

Backpropagation’s effectiveness · Digg

Backpropagation’s effectiveness

One user’s short take on backpropagation adds: “The Left has no answer to this.”

TLDR

A user offers a terse endorsement of backpropagation: “Backprop just works.” The comment continues, “The Left has no answer to this.”

Combined views

34.6K

7 Sources, first seen 44d ago

likes

Combined views

34.6K

7 Sources, first seen 44d ago

393 likes

44d ago

first seen 44d ago

393

27 comments

80 saves

23 reposts

27 comments

80 saves

23 reposts

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet

7 Sources

@stanislavfortHere's the reasoning: 1. I am conscious because I have access to the inner experience of consciousness [this is the evidentiary grounding here] 2. Other people are very similar to me, so they are also likely conscious 3. Humans came from an evolutionary line of neural nets embedded in animals, and consciousness has probably something to do with the brain 4. Along this line, the more different the animal, the less confident I am it is conscious 5. Flies live on the same line, albeit they are pretty different => there is some sense in which they could be conscious due to the above LLMs have no connection to that line. And because the only evidence we have for phenomenal consciousness is **one's own feeling", there is basically no way to extrapolate it to LLMs. So that's imho why people view the fly experiment and LLM experiment in different light, though they might not spell it out in steps.44d

@beffjezosGonna compile it to our thermodynamic computer so it's indistinguishable from artificial life, hold my La Croix40d

@fabianstelzerif you're concerned about the fruit fly (which is just the connectome and nowhere near even just a simulation of the full fruit fly brain) do NOT look up the Organoid Intelligence research around brain organoid computing ☣️📕☣️40d

@sebkrierRT @fabianstelzer: if you're concerned about the fruit fly (which is just the connectome and nowhere near even just a simulation of the ful…40d

@jachiam0I'm not especially worried that these things are conscious, or even if they are conscious that they are experiencing much meaningful badness. The amount of suffering added to the universe here is almost surely negligible. But I feel like the attitude people have towards the technology is unsettlingly irreverent. It does not inspire confidence that people would be wise if they had access to simulators that could possess consciousness.39d

@aran_nayebiI totally agree with that, and especially so once WBEs get better -- though we are fortunately far away from that, and the progress in WBEs is not nearly as fast as AI progress, so I'm optimistic that better moral frameworks will be in place by the time that is a real consideration. I also think that concordantly, building capable AI will accelerate/force us to develop scientific measures and criteria for personhood, which will hopefully be in place by the time WBE technology catches up, as I elaborate here: This notion of personhood will have to be robust to rewiring, as I note here given our NeuroAI Turing Test:39d

@teortaxesTexBackprop just works. The Left has no answer to this.38d

7 Sources

@stanislavfortHere's the reasoning: 1. I am conscious because I have access to the inner experience of consciousness [this is the evidentiary grounding here] 2. Other people are very similar to me, so they are also likely conscious 3. Humans came from an evolutionary line of neural nets embedded in animals, and consciousness has probably something to do with the brain 4. Along this line, the more different the animal, the less confident I am it is conscious 5. Flies live on the same line, albeit they are pretty different => there is some sense in which they could be conscious due to the above LLMs have no connection to that line. And because the only evidence we have for phenomenal consciousness is **one's own feeling", there is basically no way to extrapolate it to LLMs. So that's imho why people view the fly experiment and LLM experiment in different light, though they might not spell it out in steps.44d

@beffjezosGonna compile it to our thermodynamic computer so it's indistinguishable from artificial life, hold my La Croix40d

@fabianstelzerif you're concerned about the fruit fly (which is just the connectome and nowhere near even just a simulation of the full fruit fly brain) do NOT look up the Organoid Intelligence research around brain organoid computing ☣️📕☣️40d

@sebkrierRT @fabianstelzer: if you're concerned about the fruit fly (which is just the connectome and nowhere near even just a simulation of the ful…40d

@jachiam0I'm not especially worried that these things are conscious, or even if they are conscious that they are experiencing much meaningful badness. The amount of suffering added to the universe here is almost surely negligible. But I feel like the attitude people have towards the technology is unsettlingly irreverent. It does not inspire confidence that people would be wise if they had access to simulators that could possess consciousness.39d

@aran_nayebiI totally agree with that, and especially so once WBEs get better -- though we are fortunately far away from that, and the progress in WBEs is not nearly as fast as AI progress, so I'm optimistic that better moral frameworks will be in place by the time that is a real consideration. I also think that concordantly, building capable AI will accelerate/force us to develop scientific measures and criteria for personhood, which will hopefully be in place by the time WBE technology catches up, as I elaborate here: This notion of personhood will have to be robust to rewiring, as I note here given our NeuroAI Turing Test:39d

@teortaxesTexBackprop just works. The Left has no answer to this.38d

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.