“Can we trust AI?” asks Technology of Business reporter Marianne Lehnis.
Before we answer or even ask that question, we should ask :“Can we trust humans?” They can’t always articulate why they make the decisions they make, and even when they do we know that this is mostly a post hoc rationalisation for an unconscious choice. And we also know that their brains contain hundreds of bugs. Not literal bugs. Software bugs. 166 of them! They’re known as cognitive biases and they’ve been categorised.
We should be careful not to judge AI by an unrealistic standard – perfection – but instead should compare it to human decision making. Trust applies with humans too, and it is something that is built over time by a person, or a robot, proving that it is right most of the time.
It’s also worth noting that if bugs can be identified in software, they can usually be fixed. Not so in humans. At least not yet.
And humans need to sleep. That’s around 72% uptime without allowing for eating, bathroom breaks, etc.
i2O’s platform availability is 99%. And the first of our analytics modules, PRV condition monitoring, is 99% accurate. i2O can be trusted. Can you?