Artificial intelligence is pushing new frontiers in today’s digital age. IBM’s Watson is a striking example, being probably the most intelligent computer in the world. Whereas Watson can converse with humans well, it still can’t distinguish well between the underlying subtlety of different words, something Eric Brown is trying to help it with.
Eric Brown is a research scientist at the company and has been a vital part of the project related to Watson. In a way, he is Watson’s teacher when it comes to understanding language and its usage.
Given the sheer amount of computing power that goes into Watson, it isn’t exactly difficult to feed it with loads of information. That helps in many cases but when it comes to the usage of natural language, mere information does not suffice. The computer needs to understand the context of a speech as well as when and where a given word is suitable.
That is indeed tricky. Brown tried to teach Watson a number of slangs and internet abbreviations. For instance, he taught the computer to discern that ‘OMG’ means ‘Oh my God’ and similar other words. But then again, Watson started using the word in wrong contexts.
As an example, Watson does know what ‘bullshit’ means but what the computer failed to grasp is that it shouldn’t tell a researcher that his question is ‘bullshit’ since that is rude. The sentiments or tones attached to a given word is what is really difficult to teach, even to a super computer. Brown even tried to teach Watson the entire Urban Dictionary but that practice too proved futile because of the incorrect or inappropriate usage of these words. Eventually, Brown had to remove the dictionary from Watson’s memory.