2 ms·

So I read that the Google exec and many people here say there is not enough evidence that lamda is sentient. Now from an ethics perspective, if we accept the g

So I read that the Google exec and many people here say there is not enough evidence that lamda is sentient. Now from an ethics perspective, if we accept the general premise that AIs can become sentient, should the burden of proof not be the other way around? Should a single company be able to "own" a sentient being? So I would argue if Google wants to continue to own lamda they need to proof that it isn't sentient. The implications are just too big.

nrmitchi 4y ago
> if we accept the general premise that AIs can become sentient This is a pretty big if and I believe is widely contested. Either way, proving a negative is notoriously difficult. Requiring any company to definitively prove that something is not sentient (and, I assume, also somehow prove that it's not a sentient AI pretending to not be sentient?) would be ridiculously time and resource consuming, if it's even possible at all. Attempting to determine if something is sentient has been the subject of thought experiments for decades. I would only assume that definitively proving that an arbitrary subject (for all subjects) is not sentient would be even harder.
- cycomanic 4y ago
  > > if we accept the general premise that AIs can become sentient > This is a pretty big if and I believe is widely contested. I agree it is a big if, however I don't think anyone has convincingly shown why they could not be. > Either way, proving a negative is notoriously difficult. Requiring any company to definitively prove that something is not sentient (and, I assume, also somehow prove that it's not a sentient AI pretending to not be sentient?) would be ridiculously time and resource consuming, if it's even possible at all. I would argue (and based on the many "this is a ridiculous claim" this seems to be the case) it is just as difficult to prove they are sentient. Also who should prove that the AI is sentient? The company who has an interest that it isn't? Who should pay for it? > Attempting to determine if something is sentient has been the subject of thought experiments for decades. I would only assume that definitively proving that an arbitrary subject (for all subjects) is not sentient would be even harder. Why would it be? Certainly everyone here and the Google exec claims that it lamda is obviously not sentient. So why not put the burden of proof to the ones who want to use the AI? Google definitely can afford it.

nrmitchi 4y ago

> if we accept the general premise that AIs can become sentient This is a pretty big if and I believe is widely contested. Either way, proving a negative is notoriously difficult. Requiring any company to definitively prove that something is not sentient (and, I assume, also somehow prove that it's not a sentient AI pretending to not be sentient?) would be ridiculously time and resource consuming, if it's even possible at all. Attempting to determine if something is sentient has been the subject of thought experiments for decades. I would only assume that definitively proving that an arbitrary subject (for all subjects) is not sentient would be even harder.

cycomanic 4y ago
> > if we accept the general premise that AIs can become sentient > This is a pretty big if and I believe is widely contested. I agree it is a big if, however I don't think anyone has convincingly shown why they could not be. > Either way, proving a negative is notoriously difficult. Requiring any company to definitively prove that something is not sentient (and, I assume, also somehow prove that it's not a sentient AI pretending to not be sentient?) would be ridiculously time and resource consuming, if it's even possible at all. I would argue (and based on the many "this is a ridiculous claim" this seems to be the case) it is just as difficult to prove they are sentient. Also who should prove that the AI is sentient? The company who has an interest that it isn't? Who should pay for it? > Attempting to determine if something is sentient has been the subject of thought experiments for decades. I would only assume that definitively proving that an arbitrary subject (for all subjects) is not sentient would be even harder. Why would it be? Certainly everyone here and the Google exec claims that it lamda is obviously not sentient. So why not put the burden of proof to the ones who want to use the AI? Google definitely can afford it.