THE REAL FEAR OF AI
The real fear of super intelligent robots turning out to be killers might still be in the realm of fancy. There is time to react. The automation and its fall out has its own story. Nonetheless, the fear is somewhere else. In the AI systems being put in place & we blindly following it. It might lead to wrong judgements & solutions beyond the capability to correct. AI is on a pedestal beyond challenge.
Judges, auditors, doctors are already AI information as it’s from a trusted colleague. The danger is right there. Trust is killing. It’s gets it wrong, that is not important, how badly it gets in wrong is important. The systems does not fail gracefully. Husky dog was confused to be a wolf. The algorithm trained it wrong. The reverse engineering brought it out.
The Compass Criminal Sentencing Algorithm is used in 13 states of the US. Criminal recidivism is decided on that algorithm. Why should the judges trust if It exhibits a bias? Because it is a model of efficiency. The backlog has to be cleared & there’s no better way. Legally validated as it stands. But the defence is not permitted to inspect the source code. Paradox?
Should the people not inspect such source code? When they are deciding on our fate from loans to medical treatment. AI standards have to be put in place. Transparency, accountability and recourse has to be demanded from our AI systems. ISO has a formed a committee to get into it. New technology needs standards and regulation. Skepticism is the key.
Silent stimulated death cannot become the order of the day. Because somebody in a lab condition decides for us. AI is dispassionate and there is no recourse. And nobody is ready to set it right. Nobody wants to be fall guy. How far can we go? Commercially, it makes sense. We have to keep people in the loop. We have to partner with AI. We need to lead and not follow. Robots have to be is more like humans. The only thing we need to fear is our own intellectual laziness.
VALIDATE AI SYSTEMS ON SCIENTIFIC OBJECTIVITY & TRANSPARENCY.