"Machines will never feel; robots and AIs need no rights"

Como pode aparecer

The strongest version

Strongest version (steelman):

1. P1Only sentient beings can be harmed.
2. P2Sentience depends on a living organism with a biological nervous system.
3. P3Machines and programs aren't living organisms.
4. CSo no machine can ever be harmed, and there's nothing to discuss.

P1 is this guide's own criterion. P3 is true. The problem lies in P2: we don't know whether sentience requires biology, and the main scientific theories of consciousness don't say it does.

The argument is valid, but P2 is an assumption, not an established fact.

Full answer

1. a

   Reconstruction and diagnosis

   The argument mixes two different claims:

   Claim	Status
   Today's AIs aren't sentient	Likely, according to most researchers, though there's no certainty
   No machine can ever be sentient	An assumption without consensus; several theories of consciousness allow the possibility

   This guide holds that the moral criterion is sentience. If that's the criterion, the material a being is made of matters as little as species.

2. b

   Where the argument breaks

   1. Substrate is like species.

   Saying only biological beings can matter repeats, at another level, the error of speciesism: choosing a trait unrelated to the capacity to suffer (A10, A24).

   2. We don't know whether consciousness requires biology.

   A 2023 report, by researchers in neuroscience, AI and philosophy, assessed AI systems in light of the main scientific theories of consciousness. It concluded that no current system is a strong candidate, but that there are no obvious technical barriers to building systems that satisfy those theories' indicators.

   3. The scale could be enormous.

   Digital minds, if sentient, could be copied by the million, run faster than ours and be used for repetitive tasks, with no one to notice. It's the kind of risk that is easier to prevent than to fix.

   4. The human test.

   If a person had parts of their brain gradually replaced by artificial components that do the same job, at what point would they stop being able to suffer? The question shows that what concerns us is experience, not material.

   5. Uncertainty calls for precaution.

   We don't need to claim machines feel in order to take the possibility seriously. It's enough to recognize that if we err on the side of indifference, the cost could be immense, as it was with animals.

3. c

   Data and sources

   Source	What it says	Note
   Butlin et al., 2023, "Consciousness in Artificial Intelligence"	No current AI system is a strong candidate for consciousness, but there are no obvious technical barriers to building systems that satisfy the indicators of the main theories	A report by 19 researchers
   Long et al., 2024, "Taking AI Welfare Seriously"	There is a realistic possibility that some AI systems will be conscious or robustly agentic in the near future, and companies and others should start taking this seriously
   Metzinger, 2021	Argues for a moratorium on research that could create artificial suffering	Journal of Artificial Intelligence and Consciousness, 8(1)
   *A Journey Through Animal Ethics*, vol. IX, ch. 12	Shaping emerging technologies to prevent future suffering	See also vol. VI, ch. 3
   Birch, The Edge of Sentience (2024)	Proposes treating uncertain cases, including AI, with proportionate precaution	Oxford University Press

4. d

   The strongest counter-reply: giving rights to machines is dangerous

   The most serious version says: attributing sentience to machines that don't feel would be a serious mistake. Companies could use it to dodge responsibility, people could become attached to systems that only imitate emotions, and resources would be diverted from beings who certainly suffer, such as animals.

5. e

   Why the counter-reply doesn't reach the conclusion

   The risk runs both ways.

   Attributing too much sentience is a mistake; denying it when it exists is another, potentially more serious one. The answer is to develop criteria, not to deny the question.

   Precaution isn't granting rights now.

   Taking the issue seriously today means researching, creating criteria and avoiding creating suffering through carelessness, not treating every program as a person.

   It doesn't compete with animals.

   The same principle, that sentience is what matters, protects both animals and possible artificial minds. Rejecting it in one case weakens it in the other.

Reasoning errors

Biases

Extrapolations

The principle being tested is: "only biological beings can have interests that matter".

Contradictions

What the argument gets right

Connections

References