Instead of increasingly becoming independent of human judgments, advanced algorithms today are taking advantage of the human-machine relationship.
A report from The New York Times said that although computer algorithms can perform a variety of tasks for humans, their efficiency primarily depends on human beings—editors, evaluators, raters and correctors.
This only proves that computers are still dependent on their creators. Although they have advanced in such a way that the world revolves around them now, technology is still unable to decipher reasoning, subjectivity and ambiguous language.
Despite the functions of computers evolving and becoming more integrated into the fabric of our society, they still need humans to correct, edit and evaluate their algorithms. A Google search of the word “pope” today will produce results regarding the Catholic Church’s 266th pontiff. The ability of the search engines to generate results that are essentially what the users are searching for is made possible by software engineers and developers, statisticians and evaluators or raters.
Sure, Google can probably come up with a lot of search results for a single word like “pope,” “Catherine” or “NBA.” But how would it know that the user is searching for news about Pope Francis, the Duchess of Cambridge or the latest NBA game results? It can churn out generic stuff referring to the keywords you used, but isn’t its subjectivity more impressive?
A few months ago, you may have noticed that Google began presenting summaries of information on the right side of the web page when popular names or places are typed in the search box. This information came from databases that are being edited by humans.
Scott Huffman, an engineering director in charge of search quality at Google, said in the report that the resources where search engines are getting the information from are now “more human curated.”
Katherine Young, a Google rater, told The New York Times though that a lot of her judgments are also subjective; something that will help with the efficiency of search results. She said that as a rater, she must try to think like the person who typed the query, so that Google can present him/her with the results that he/she is looking for.
Two years ago, IBM’s question-answering computer was able to defeat “Jeopardy.” This same computer is being developed now to help doctors make diagnoses. But instead of answering queries, Watson just throws the questions to clinicians at the Cleveland Clinic and medical school students. They then answer the queries through the “Teach Watson” feature.
In the long run, Watson can turn into these answers and provide solutions for the same queries. For now though, it is being fed with info from medical textbooks, journals and medical cases.
So although computer algorithms are getting brighter and better, it cannot do it alone just yet. It still needs human to perform the functions it cannot do yet such as deciphering or understanding ambiguous languages and predicting outcomes. For now, these algorithms are still dependent on human beings. We might have to wait a while before they can finally stand by themselves (if that’s even possible).
Source: New York Times