So I've been digging into this topic for days now. I've gone through his paper and also some threads here, which I found were very helpful. I came up with these premises and following conclusions in order to refute Searle's argument, and support that computers can think and have semantic processing.

P1) Just as Searle says, AI system is strictly syntactical, following rules & symbols.

P2) AI system is a black box. There are controversies on how while AI is intelligent, the human creators don’t exactly not know what’s going on inside the system (ie. What they are actually learning, or when there is bug, and etc).

P3) AI system is more complex than calculators. Calculators are simple system that can be boiled down just to electronic circuits manipulating numbers, but AI system are definitely more complex.

P4) AI system is a mimicry of the human brain (ie. neural network is an example of multiple neurons, and I support the connectionism refutation that multiple neurons would eventually as a whole could give semantic experience.

C1) AI is like a puppy robot. Even though AI system is strictly defined syntactically (robot), we do not exactly know what’s going on inside (puppy). Thus, it must have its own semantics possibly developed.

C2) Just as AI is superior to calculators, brain is more complex and superior system to AI that our current technology hasn’t achieved yet. Comparing human brain to AI techniques is not adequate.

I wanted to test the soundness of my statements, and any comments are welcome. Thank you.

  • The title is off, it is impossible to "refute" an experiment, even a thought experiment. So I am not sure what the question is exactly? To check if the argument is valid, are Ps supposed to be premises and Cs conclusions? In that case I do not see how "it must have its own semantics" follows from "we do not exactly know", or where "comparing human brain to AI is not adequate" comes from, computers work on the same principles of von Neumann architecture as calculators. – Conifold Feb 25 '17 at 0:18

To my mind P4 and P2 contradict one another. We may not be able to comprehend the details all at once, but if we understand the design, this is not a black box.

Bugs are not evidence of independent thought. Neither is sheer scale. We do not know down to the molecule what is going on in a box of paper clips, that does not mean it thinks.

So C1 is not really supported. We know the semantics the robot is intended to have. Failure to have them, or the fact that what we set in motion overwhelms us in detail via intrinsic chaos, is not evidence of its intelligence. A wind vortex overwhelms us in detail via intrinsic chaos, and remains without thought.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.