Today’s artificial intelligence may be able to detect if a patient needs surgery or when someone is committing fraud, but humans are often left in the dark on how these machines reach those conclusions. So, Google’s Been Kim has developed a kind of human-machine translator to fill that gap. The tool, called TCAV, allows researchers to ask AI systems to identify the most important pieces of information that led to its conclusions.
Passionate about learning (human & machine) | Agile believer | Socrates' dialectic rules
Perhaps in the future machines can help us understand how our own decisioning process works? https://lnkd.in/gvJckPR
Been Kim, a research scientist at Google Brain, believes that we should expect nothing less from artificial intelligence. As a specialist in “interpretable” machine learning, she wants to build AI software that can explain itself to anyone.
"I don’t tell you, “Oh, look at pixels 100 to 200, the RGB values are 0.2 and 0.3.” I say, “There’s a picture of a dog with really puffy hair.” That’s how humans communicate — with concepts."
A huge factor in humans being able to trust black box AI is assuring ourselves that the AI bases its decisions on the same features and concepts that we would use.
"Further, it can be difficult to discern how a mathematical model trained by deep learning arrives at a particular prediction, recommendation, or decision. A black box, even one that does what it’s supposed to, may have limited utility, especially where the predictions or decisions impact society and hold ramifications that can affect individual well-being." Algorithms are increasingly helping us make life-changing decisions, hence transparency is more important than ever. Always know the Why and the How to the best of your ability. Or get a 2nd or 3rd opinion. Asking for help is a sign of humility, not weakness. . . . . . . #datascience #ai #machinelearning #marketing #socialmediamarketing #socialmedia
Interpretable AI has been discussed a lot especially in the last year. As we start 2019, here is an article on a new approach to understanding how machines think. It is well explained, has some good examples and now I'm curious enough to dig further and read the published papers on the topic. https://lnkd.in/eNNQTkx
The complex neural networks at the core of machine learning make it difficult - perhaps impossible - to exactly pinpoint how AI systems draw conclusions and make decisions.
It'll be cool to see the "WHY" behind what machine learning is doing to surface answers. #machinelearning
A well written piece on the challenges of black-box AI https://lnkd.in/efu3ggr #wellness #artificialintelligence #ethics