2459 - More Trustworthy GenAI and ML with IBM Decisions 101824.pdf

1、24592459More Trustworthy GenAI and ML with IBM DecisionsDEDICATED TO DECISIONS FIRST James Executive Partner Decision Management Solutions More than 20 years championing Decision Management,business rules,analytics and decision modeling Frequently cited expert and author Lenny Program Director IBM 2

2、0+years working in Decision Management(consulting,tech sales,sales,product management)with IBM across the globe 2024 Decision Management Solutions 2IntroductionsWhat Builds Trust?HONESTYCONSISTENCYTRANSPARENCYCOMMUNICATIONTrust is an Issue for GenAI And for Machine Learning more generallyBias in mod

3、els resulting from bias in the underlying data undermines honesty by perpetuating and exaggerating historical bias.ML and AI models are often“Black Boxes”,opaque and hard to follow.This eliminates needed transparencyHallucinations in AI and the probabilistic nature of ML models undermine communicati

4、on by creating inconsistency in answers.Technical complexity and a tendency to rely on self-justification can undermines the ability to communicate exactly what an ML or AI model is doing.Generative AI is not Artificial Intelligence(AI)AI is going to change the world AI Adoption Build trust and succ

5、eed with AI and ML Mix and Match To Increase Impact WrapAgendaGenAI is not Artificial IntelligenceAny technology that can be used to make a machine act“intelligently”Machine LearningExpert SystemDecision EngineDeep LearningConstraint Based OptimizationImage RecognitionNatural Language RecognitionDat

6、a MiningGenAILarge Language ModelPredictive AnalyticsBusiness RuleDecision ModelRegression ModelNeural NetworkScorecardDecision TableArtificial IntelligenceAs Gartner saysAI Does not revolve around GenAIOur experience is that three technologies in particular complement GenAIWatson MLIBM Machine Lear