Do we know how to build machines that understand language? Though some good attempts are out there, not yet. This Tech Review Article article does a good job of looking at the history of the challenge. Have spent some time in those years addressing the leveraging of language.
Systems like Cortana, Siri and Echo seem to understand language well enough to get some clever things done, but you soon find out that the intelligence at play is still thin. As long as it matches some template, or links to a search in some way, or has a fairly predictable formulaic answer ... it seems to work. But the result is not always satisfactory, and you find yourself tailoring answers or solutions to your needs. And worse, the system does not learn from the conversation. And in later interaction does not learn to focus on your needs.
Like talking to someone who is very adept at calculating in their head, or looking up facts, but has limited common sense. And is not considering the varying context of a conversation. Results can still be useful, just like searches are, but we frequently have to adapt, and often re-ask questions.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment