The AI Morality Divide

There's been a lot of commentary on the value as well as the luggage that GenAI brings to our internet always-on always-connected society. From expert perspectives in the field of modeling, to opinions from lay-persons, there seems to be a growing divide between whether GenAI/LLMs are good or bad for us.

The Good

On the positive side, AI models have found exciting new scientific uses in discovering protein combinations we never could have calculated on our own, determining early indicators of cancer, and addressing other truly complex problems in math and linguistics.

I envision advances in AI may finally decode communication protocols in highly intelligent species such as primates, dolphins, horses and elephants.

The Bad

On the negative side, we can see that upcoming U.S. elections will be rife with disinformation using AI-generated images, audio and video. It will be impossible for the American public to discern any information about the candidates to be valid. (That should be extremely troubling for all countries.)

Beyond elections, I am suspect of any news or videos on almost any subject. It used to be that pictures and videos were provided as definitive proof. We can no longer trust any media.

The Ugly

The point if this article is to show that there are AI use cases with no clear positive or negative outcome. In fact, the use case I present here is extremely uncomfortable to even think about. But this example is needed to start a different conversation.

In a similar vane,

• Does "Grand Theft Auto" contribute to the rise in carjackings, car thefts, or even just bad driving?
• Do first-person shooter games such as "Call of Duty" create a training ground for active shooters?
• Here is the ugliest juxtaposition: Is AI-generated child pornography a positive use case if it means no children are harmed? In considering the physical safety of children over the psychological reinforcement of deviants, does this AI use case provide an alternative?

Clearly this is not something anyone would want to consider, and the use case itself does not address the legalities of the situation, nor the propensity for allowing unsavory perpetrators to exist.

I don't know the answer, but it is a scenario that weighs heavily on me.

I also want to admit that each of these scenarios are exacerbated with the darker depths of social media that proliferate and even promote abnormal behaviors.

Update: in speaking with many [active & former] law enforcement peers, they would choose to take actions that reduce the deviance than the alternative given. The trade-off, to paraphrase, is to work towards saving hundreds of lives in the long run.

Summary

Perhaps it is human nature to be divisive, to find internal solace in taking only one side. However, once you have seen one ugly example, can we start reconsidering any AI outcome beyond being only positive or negative?