3 ms·

Exactly. Anthropomorphic language is such a useful and normal way of thinking and expressing ourselves. If you say "the classifier thinks this roadsign is a tr

Exactly. Anthropomorphic language is such a useful and normal way of thinking and expressing ourselves. If you say "the classifier thinks this roadsign is a truck" everyone knows you mean "the classifier has labeled this roadsign as a probably a truck with a high confidence". No-one seriously would take you to mean the classifier thinks in the same way a person thinks any more than if you say "The toaster wants to pop up but it's jammed," they would take you to mean the toaster has a yearning like unrequited love. If we remove our ability to use language in this way we give up on a mental model that is incredibly useful for simplifying lots of explanations and reduce ourselves to a very clumsy literalism instead. As long as we understand the limitations of this model and are clear that it's not literally what is going on it's very useful.

qsort 6y ago
My problem is not with natural language in and of itself, it's obvious that natural language is extremely useful. I'm arguing one should be extremely careful, however. Natural language doesn't have deduction. It doesn't have definitions. It doesnt't have ways to restrict us from making vague statements. It is subject to non-unique interpretations. This is particularly relevant when technicians and scientists speak to the general public. Let me quote one of your examples: > "the classifier thinks this roadsign is a truck" everyone knows you mean "the classifier has labeled this roadsign as a probably a truck with a high confidence". The fact that the classification is probabilistic is obvious to me and to you, but nothing in the natural wording suggests that, and it's a pretty important caveat. It's not enough for us to understand the limitations of natural language, everyone in the conversation has to, for it to be a useful tool.
- rusk 6y ago
  “The classifier has guessed that ...”
  - qsort 6y ago
    The fact we're arguing what exact wording to use is kind of the point, isn't it?
    - rusk 6y ago
      Not arguing, just suggesting an alternative phrasing that might better convey the intent. When people say thinks in this sense they mean determines to a certain level of confidence, rather than modelling consciousness.
- Talanes 6y ago
  "Thinks" in that context is inherently probabilistic. No one uses the phrase "I think..." to literally state that they have a thought process going through their head, they do it to clarify that the following statement is based on their own conclusions rather than objective reality.
  - qsort 6y ago
    I don't think that is the case.
  - TheOtherHobbes 6y ago
    No one except Descartes: "I think therefore I am." But all this does is underline that natural language is a vague mesh of concept categories anchored to words with multiple overloaded meanings. One of the reasons philosophy exists is because sometimes it attempts to untangle the overloading. The argument about AI is no different, but with the complication that the marketing concept mesh has different requirements to the research concept mesh, and both are different to "Make this work by next Tuesday" mesh. Google used to have an ancestor of word2vec which tried - very crudely - to map some of these meshes. Predictably it was much better at mapping word meshes than concept meshes, but it was still an interesting approach. The underlying problem is that we literally don't have a clear representation system for these concept meshes. Words are overloaded, code and math are too specific in their different ways. The various shades of ML are still too concrete. So we're going to keep having these debates until someone invents a representation system for natural language concept maps that isn't based on verbal approximations.

qsort 6y ago

My problem is not with natural language in and of itself, it's obvious that natural language is extremely useful. I'm arguing one should be extremely careful, however. Natural language doesn't have deduction. It doesn't have definitions. It doesnt't have ways to restrict us from making vague statements. It is subject to non-unique interpretations. This is particularly relevant when technicians and scientists speak to the general public. Let me quote one of your examples: > "the classifier thinks this roadsign is a truck" everyone knows you mean "the classifier has labeled this roadsign as a probably a truck with a high confidence". The fact that the classification is probabilistic is obvious to me and to you, but nothing in the natural wording suggests that, and it's a pretty important caveat. It's not enough for us to understand the limitations of natural language, everyone in the conversation has to, for it to be a useful tool.

rusk 6y ago
“The classifier has guessed that ...”
- qsort 6y ago
  The fact we're arguing what exact wording to use is kind of the point, isn't it?
  - rusk 6y ago
    Not arguing, just suggesting an alternative phrasing that might better convey the intent. When people say thinks in this sense they mean determines to a certain level of confidence, rather than modelling consciousness.
Talanes 6y ago
"Thinks" in that context is inherently probabilistic. No one uses the phrase "I think..." to literally state that they have a thought process going through their head, they do it to clarify that the following statement is based on their own conclusions rather than objective reality.
- qsort 6y ago
  I don't think that is the case.
- TheOtherHobbes 6y ago
  No one except Descartes: "I think therefore I am." But all this does is underline that natural language is a vague mesh of concept categories anchored to words with multiple overloaded meanings. One of the reasons philosophy exists is because sometimes it attempts to untangle the overloading. The argument about AI is no different, but with the complication that the marketing concept mesh has different requirements to the research concept mesh, and both are different to "Make this work by next Tuesday" mesh. Google used to have an ancestor of word2vec which tried - very crudely - to map some of these meshes. Predictably it was much better at mapping word meshes than concept meshes, but it was still an interesting approach. The underlying problem is that we literally don't have a clear representation system for these concept meshes. Words are overloaded, code and math are too specific in their different ways. The various shades of ML are still too concrete. So we're going to keep having these debates until someone invents a representation system for natural language concept maps that isn't based on verbal approximations.