There is no other option for Mr.Mark Zuckerberg, than to be optimistic. Optimistic about everything including, AI would salvage his dooming company.
Sometimes, when you have been a so called ‘pioneer’, you can very escape by making innocuous predictions. Those which others cant help to do anything else, except nod their heads. This is what Mr.Zuckerberg did at US Congress hearing, saying that AI would solve the ‘fake news’ problem in 5 to 10 years horizon.
Lets take the perspective on this:
Today’s A.I. operates at the “keyword” level. Flagging word patterns and looking for statistical correlations among them and their sources. While this is OK to some sense, but this is like trying to paint the twin, black and white. Fake stories usually, if not always try to make a causal connection, which does not exist. And this where the current machine learning techniques fail.
Understanding causal relationships requires a lot of cognition, which is not available to current AI systems. Like identifying what to do next, which pieces of information to look; to verify the implied relationships. And beyond this, when still the doubt persists, then to seek different view points from diverse sources.
In such scenario, verifying a simple fact may prove arduous to a AI system, let alone, prove a story being fake, which could consist of multitude of facts.
Most of the current news comprehension AI system, can answer only binary questions or classifications.
To get to where Mr. Zuckerberg wants to go will require the development of a fundamentally new A.I. paradigm, one in which the goal is not to detect statistical trends but to uncover ideas and the relations between them. Only then will such promises about A.I. become reality, rather than science fiction (source).
Fake news is a problem to be solved only through human cognition, when it is generated by artificial intelligence. The mis-match in scale is what is going to prove fatal. For whom, the time will tell…
MIT shows how to tackle fake news using AI and ML