The AI generated image to the left is supposed to show professionals at a whiteboard arguing.  The AI took plenty of liberties with the image, although I generally agree with them.  Still, the women are wearing completely impractical shoes for an office setting and the bearded man needs better fitting clothes.  And whatever you do, don’t count the fingers of the people in the image.

I’m not trying to say that Generative AI is “bad”, but I do want to caution those folks who insist that AI is the beginning of the end of human ingenuity that the image presented here, as good as it is, does not pass muster for use in a business setting.  I wouldn’t put it into a presentation for senior management.

Of course, I can work to refine it, and there’s the rub.  Generative AI is derivative.  Systems that generate images like these do not “know” anything.  They are putting together bits of data from multiple places and spitting out something without regard to what the image “means” to a person.  

While this impersonal reality of AI strikes fear into many, I challenge that we need another layer.  We need another AI  to govern the generative AI.  The governance modules are clearly not at work in the image above, but it is no more difficult to add a governance model than it is to create the generative model above.

What would that governance model do?  It would argue with the generative AI.  Constrain it.  Improve it.  Perhaps create a conversation with the user to understand how it is received.

I suggest that Asimov’s three laws of robotics stem not from the Generative side of AI but from the Governance side, the future module that we don’t have fully created yet.  

So before we create a panic about AI, let’s recognize that the model of true intelligence requires the ability to self govern.  And we are not there yet.