Key Arguments

• He explains that fringe consciousness "takes account of cues in the context, and probably some possible parsings and meanings, all of which would have to be made explicit in the output of a machine," highlighting the difference between tacit human awareness and explicit machine representation.
• Our 'sense of the situation' allows us to exclude most possibilities before they ever become candidates for consideration; they simply never arise as options, rather than being enumerated and rejected.
• He proposes to call "the ability to narrow down the spectrum of possible meanings by ignoring what, out of context, would be ambiguities, 'ambiguity tolerance.'"
• This capacity presupposes the first form of information processing (fringe consciousness) and explains why humans can use and understand sentences in familiar situations without needing to formalize away all ambiguity.
• Since people use and grasp language against a rich background of practical familiarity, Dreyfus suggests that the only route to machines that could understand and translate real utterances would be to make them learn about the world—an idea already suspected by Turing.
• He cites Bar‑Hillel’s claim that machines that cannot 'learn, in a sophisticated sense' will never consistently achieve high‑quality translation, showing that even optimistic workers recognize that brute formalization is insufficient and appeal to richer world‑learning.

Source Quotes

Fringe consciousness takes account of cues in the context, and probably some possible parsings and meanings, all of which would have to be made explicit in the output of a machine. Our sense of the situation, however, allows us to exclude most possibilities without their ever coming up for consideration. We shall call the ability to narrow down the spectrum of possible meanings by ignoring what, out of context, would be ambiguities, "ambiguity tolerance."

Our sense of the situation, however, allows us to exclude most possibilities without their ever coming up for consideration. We shall call the ability to narrow down the spectrum of possible meanings by ignoring what, out of context, would be ambiguities, "ambiguity tolerance." Since a human being uses and understands sentences in familiar situations, the only way to make a computer that can understand actual utterances and translate a natural language may well be, as Turing suspected, to program it to learn about the world.

43* Our ability to use a global context to reduce ambiguity sufficiently without having to formalize (that is, eliminate ambiguity altogether) reveals a second fundamental form of human "information processing," which presupposes the first. Fringe consciousness takes account of cues in the context, and probably some possible parsings and meanings, all of which would have to be made explicit in the output of a machine. Our sense of the situation, however, allows us to exclude most possibilities without their ever coming up for consideration.

We shall call the ability to narrow down the spectrum of possible meanings by ignoring what, out of context, would be ambiguities, "ambiguity tolerance." Since a human being uses and understands sentences in familiar situations, the only way to make a computer that can understand actual utterances and translate a natural language may well be, as Turing suspected, to program it to learn about the world. Bar-Hillel remarks: "I do not believe that machines whose programs do not enable them to learn, in a sophisticated sense of this word, will ever be able to consistently produce high-quality translations."44 When occasionally artificial intelligence enthusiasts admit the difficulties confronting present techniques, the appeal to learning is a favorite panacea.

Since a human being uses and understands sentences in familiar situations, the only way to make a computer that can understand actual utterances and translate a natural language may well be, as Turing suspected, to program it to learn about the world. Bar-Hillel remarks: "I do not believe that machines whose programs do not enable them to learn, in a sophisticated sense of this word, will ever be able to consistently produce high-quality translations."44 When occasionally artificial intelligence enthusiasts admit the difficulties confronting present techniques, the appeal to learning is a favorite panacea. Seymour Papert of M.I.T., for example, has recently claimed that one cannot expect machines to perform like adults unless they are first taught, and that what is needed is a machine with the child's ability to learn.

Key Concepts

• Our sense of the situation, however, allows us to exclude most possibilities without their ever coming up for consideration.
• We shall call the ability to narrow down the spectrum of possible meanings by ignoring what, out of context, would be ambiguities, "ambiguity tolerance."
• Fringe consciousness takes account of cues in the context, and probably some possible parsings and meanings, all of which would have to be made explicit in the output of a machine.
• Since a human being uses and understands sentences in familiar situations, the only way to make a computer that can understand actual utterances and translate a natural language may well be, as Turing suspected, to program it to learn about the world.
• "I do not believe that machines whose programs do not enable them to learn, in a sophisticated sense of this word, will ever be able to consistently produce high-quality translations."44

Context

Still within 'Ambiguity Tolerance vs. Context-Free Precision', Dreyfus explicitly names and characterizes ambiguity tolerance as a distinct, irreducibly contextual human capacity and connects it to the idea that any machine approaching human‑level language understanding must somehow acquire situated world knowledge.