Problem With Processing: Humans vs Machines

Tap to Read ➤ Craig Middleton

As machines get even more and more smarter as well as the technology behind these machines gets increasingly powerful, we use them more and more in our daily lives.

As a result, machines and humans interact more than ever before, and in more complex ways. It isn't just a matter of pushing a button or flipping a switch, many people speak to their machines, whether they're asking a phone to set a timer or sending a text hands-free, using speech to text functions.

It's interesting to examine the complexity of human language and expression, looking specifically at the difficulties that machines will likely have when processing human communication.

Language

Language is the first and most obvious component of human communication. As a matter of fact, humans have been communicating with machines for years using language. The difference is that humans generally communicate with machines in the languages they used to program those machines.

In layman's terms: humans have been speaking to machines in the machines' language. Now, with machines learning and NLP and other algorithms, machines are starting to be able to communicate with humans in human language.

Structure and Meaning

If you're wondering what is NLP , it stands for Natural Language Processing, a system by which computers use algorithms to analyze human language. Natural language processing, among other roles, allows word processors to analyze and correct grammar by looking at the letters and the words in a sentence, and also the grammatical structure and the meaning.

All of this brings computers closer than ever to what we might understand as comprehension of human communication. While NLP has been a goal of computing for decades, it is still a developing field. The next challenge is going to be bringing in the subtleties and nuance of human communication.

Tone

Tone of voice plays a huge role in human communication. Think about the phrase "let's talk," for example. Hearing it spoken in an enthusiastic way, perhaps from an old friend, can convey one meaning. Hearing it said in a low, serious tone, from a boss or a spouse, carries a very different set of connotations.

That's a two word sentence, with different meanings based entirely on tone. Tone is hard to measure, and harder to program. Machines are only just learning how to interpret tone, and that challenge isn't helped by the fact that the use of tone to convey meaning varies between languages, and even between dialects of the same language.

Visual Cues

Certain AI programs are beginning to mimic visual cues, and even recognize and process them. Many of you will be familiar with videos of robots with human-like faces miming human expressions, or even the way that video game characters respond to the different choices that player characters can make.

Visual cues are a huge part of human communication, from facial expressions to hand gestures and body language. Visual cues are important indicators when you're speaking to a human, but hard for a computer to pick up and process.

Social Cues

Beyond body language are unspoken, un-signaled social cues that rely on prior knowledge of a social circle. Saying "Allison said so" has very different connotations depending on your prior knowledge of Allison. Is Allison a reliable person or is she a compulsive liar? It's easy to miss social cues if you're not in someone's social circle.

It's even easier to miss social cues if you're already struggling to interpret tone and visual cues, as many AI programs do. That said, theoretically at least, AI programs have a much broader base of knowledge to work from.

Hypothetically, a machine connected to the internet could assess the social connections and cues underlying a statement, perhaps more comprehensively than their human counterpart.

Doubtless, as technology continues its inevitable march forward, humans will interact with machines in increasingly complex ways. As a result, machines will have to become increasingly adept at analyzing human speech and all the nuances of meaning expressed by it.