>>Did he send these specs to you on April 1st?

LO. No, believe it or not, he's serious. He's paying me to do the research. I'm trying to find a nice way to tell him.




>>I think the whole thing is wishful thinking. What I need from you guys is input. Is what he wants possible? Under what circumstances?
>
>Think about what it would take to do this manually.
>
>Set up a bank of (Indian?) tech support guys monitoring 5 screens at a time to follow along silently while your user works.
>You would need a lot guys, and they would have to change up who they are watching periodicallyto prevent devastating boredom.
>Eventually, one of them notices a user fiddling around.
>He has to decide if the user is "in trouble" or just normal working.
>
>(All the steps to this point require human understanding and intelligence, not possible with any kind of AI)
>
>If you think about it, even a human could not figure it out from just watching.
>He would need to have the user talk to him about what he is trying to do.
>
>This decision also requires understanding this particular user:
> What is his ability in this task?
> What is his capacity for frustration?
> Some users are happy to figure stuff out on their own, some would be very annoyed to have Clippy pop up and offer help.
>
>Assuming you could actually decide if the user is "in trouble", now you need to offer reasonable assistance.
>
>Your theoretical help desk guy cannot be expert on every nuance of Word, Excel, FoxPro, C#, Powerpoint, etc.
>So, now he has to know who can help him with this user.
>
>This project is deeply into the realm of "not even theoretically possible"
>You can't make software that can think and reason - not possible.
>
>These specs seem to assume that software can think and reason, and do so much better than an expert human.
>
>The only person on the planet who can know if that user is "in trouble" is that user.
>He should just call the help desk when he needs help ;)
>
>Did he send these specs to you on April 1st?