Against Strong AI

Against Strong AI

Hi, I'm Tim Tyler - and today I will be discussing the term "Strong AI" - and why you should probably avoid using it.

Origin of the term

"Strong AI" is terminology originally invented by the machine intelligence sceptic John Searle.

He used "The weak AI hypothesis" to refer to the idea that machine intelligences only act as though they think and have minds.

He used "The strong AI hypothesis" to refer to the idea that machine intelligences actually think and have minds.

The rhetorical point was that the strong AI hypothesis was a made claims that went beyond the weak AI hypothesis - and was indistinguishable from it experimentally - therefore you should prefer the weak AI hypothesis, because it is more conservative - and does not make additional claims that are not justified by the experimental evidence.

Of course, this distinction is an idiotic one - but that's not the point of this video.


What happened next is that people overheard experts discussing the strong AI hypothesis. Naturally they were not able to make much sense of what they heard: if it walks like a duck and quacks like a duck then why not just call it a duck?

However, they could tell that the machine intelligence supporters were saying that the machines were "strong" and that the sceptics were saying that they were "weak". The natural conclusion was that the terms "strong" and "weak" had something to do with the capabilities of the machine intelligences - rather than refering to the strengths of hypotheses, as originally intended.

Machine intelligences do not necessarily have muscles - so calling them "strong" makes little sense - except that "strong" could be regarded as being part of an analogy with the term "powerful".

However, this new usage made just enough sense to catch on - and it eventually eclipsed John Searle's original intended usage.

I think the terminology is as stupid as its origins suggest. Calling an AI "intelligent", "smart" or "powerful" would make sense - but the term "strong" is just silly.

The term "Strong AI" is practically useless in John Searle's original sense - because the concept it refers to is a pointless one.

It is practically useless in the context of a powerful machine intelligence, because of the ambiguity with John Searle's terminology - and because "strong" is a dreadful synonym for "smart".

I think that the best thing to do is avoid the terminology altogether.

One of the main effects of using it is to give the impression to people such as myself that you do not really know what you are talking about.



  1. Wikipedia - Strong AI;

Tim Tyler | Contact |