The AI doesnt understand the former, only the latter.
Do you possibly mean “The AI evangelists” or something similar?
Like, I could totally understand it in the “software will also include the biases of those who wrote it” kind of way (a la Amazon’s failed attempt at automating job candidate search). If the only incentive you’re given as a programmer is “make it make money”, then yeah, your AI is going to bias towards that end.
I’m not actually asking for good faith answers to these questions. Asking seems the best way to illustrate the concept.
Does the programmer fully control the extents of human meaning as the computation progresses, or is the value in leveraging ignorance of what the software will choose?
Shall we replace our judges with an AI?
Does the software understand the human meaning in what it does?
The problem with the majority of the AI projects I’ve seen (in rejecting many offers) is that the stakeholders believe they’ve significantly more influence over the human meaning of the results than exists in the quality and nature of the data they’ve access to. A scope of data limits a resultant scope of information, which limits a scope of meaning. Stakeholders want to break the rules with “AI voodoo”. Then, someone comes along and sells the suckers their snake oil.
Do you possibly mean “The AI evangelists” or something similar?
Like, I could totally understand it in the “software will also include the biases of those who wrote it” kind of way (a la Amazon’s failed attempt at automating job candidate search). If the only incentive you’re given as a programmer is “make it make money”, then yeah, your AI is going to bias towards that end.
Just couldn’t tell on first reading
I’m not actually asking for good faith answers to these questions. Asking seems the best way to illustrate the concept.
Does the programmer fully control the extents of human meaning as the computation progresses, or is the value in leveraging ignorance of what the software will choose?
Shall we replace our judges with an AI?
Does the software understand the human meaning in what it does?
The problem with the majority of the AI projects I’ve seen (in rejecting many offers) is that the stakeholders believe they’ve significantly more influence over the human meaning of the results than exists in the quality and nature of the data they’ve access to. A scope of data limits a resultant scope of information, which limits a scope of meaning. Stakeholders want to break the rules with “AI voodoo”. Then, someone comes along and sells the suckers their snake oil.
deleted by creator
Oooh la la