On the question of commonality (i.e., AI machines, while using common data sets for training when presented with a common input don’t give the same output, i.e., they don’t arrive at common conclusions), and trust (experts are inherently skeptical of the abilities of a machine) are pervasive topics in today’s data science/AI discussions. Let’s explore these points through the prism of our human-to-human interactions.
First Commonality. If I give two analysts the same data point and trends will they always arrive at the same decision regarding the appropriate next step? No. Why? Because how they have been trained to date differs (even if the actual training data is common). This training imparts a fingerprint on the individual and causes them to view data inputs in different ways. Even if two humans shadow each other from birth in regards to education and social interactions, if their training programs vary at all they will inherently interpret new inputs differently over time. Every machine is trained by a different group of data scientists and programmers. Thus it is highly unlikely, if not impossible, for any two machines to arrive at a similar conclusion. Like our parents, teachers, bosses/co-workers, and spouses, every data scientist and programmer imparts influence on the machine that effects the resulting decisions. This problem is effective one of a chaotic system; butterfly effects are present.
Second Trust. Doesn’t it take us a long time to earn the trust of others? As our passion and knowledge for something grow the steps needed to trust somebody or something grows exponentially. We are all experts in some small aspect of our lives, how easily would you trust another in handling these affairs? I would argue it would take months if not years for that trust to be earned. And by the meaning of earned, we can say that it has proven effective, and has been explained to our satisfaction (by analogy, or analysis, or other persuasive means.) Why do we expect anything different from machines? If you have one bad experience with machines, don’t let it bias your entire opinion on the significance machines are having and will continue to have in the world. Give it time, and you will see that machines will earn your trust.
Like the snowflake, no two humans and no two machines are the same. Our ecosystems mold us in different ways and lead us to be unique. Like the stray cat, it will take time for us to trust machines fully. Our expertise makes us leery of this new entrant into our lives. If we embrace the ecosystem and accept that machines can create value for us, then over time an affinity for the machines will form.