Gary Marcus

Are you saying your robot lawn mower has no intention of mowing the lawn either? Wait are you saying my robot lawn mower has intentions?

Not I. @NoelSharkey might have...insofar as he asked the question: "Did the robot lawn mower decide to come and mow your lawn?" As for teenagers, the jury is still out regarding intention when it comes to matters of household chores.

Pretty sure @NoelSharkey's ques was rhetorical & he doesnt think robot mowers have intentions. Your analogy is faulty by comparing one sys capable of intentions to another without it around an issue that concerns that very intentionality.

It may be perfectly sensible to describe a robot lawnmower as deciding to mow your lawn, if the robot lawnmower considered a variety of alternative possibilities first, and then chose one.

It doesn't cconsider. It follows the routine. Sensory input/perceive-categorize-on-the-background-of-the-training-data/NN >> act. It's Math, iterated, Dylan, plus energy ON/OFF to start mowing.

But it makes sense to describe the lawnmower in this way if it really does have a lot of options to choose between. Just like it makes sense to describe humans in this way, although of course humans too are just math, iterated, plus energy

I know that some think that the map is the territory. It isn't. Plus: perception != understanding. See: Psychology 101 @GaryMarcus has questioned the hubris of taking AI models for true abstractions of human neural processes.

Actually, @GaryMarcus questioned a very specific thing - whether neural nets are good abstractions of human nervous systems. On the broader question of AI models in general, he certainly does think that some of them *are* good models for human mental processes.