Click here to download PDF copy of presentation

Speaker: John Mark Agosta (Applied A.I Researcher)

Abstract:

Is it more valuable to make millions of 1 dollar decisions or one 1 million dollar decision? Automating decisions in software opens a new frontier in Decision Science. It is another proving ground for Decision Quality (DQ).

Don’t get confused by what “AI” was and what has it now become. AI’s origins have common roots with DQ in the pursuit of rationality. The siren song of current generative AI language models and their appearance of “common sense” do not replicate rational decision making. This talk explains how the value modelling familiar to Decision Analysts can be integrated to complement Data Science’s predictive modelling, to solve the so-called “alignment” problem. Then speculatively, I broach how this new world of automation raises an interesting moral question in light of Newcomb’s paradox.