Caster’s Baby (Part 1) {Your Baby}
Imagine a world where you can converse with a machine without feeling crazy, the machine completely gets you and understands where you are coming from. I mean it can empathize with you.
I’m talking of a world where an applicant-filtering software goes through a resume and does not disqualify a candidate for just grammatical errors but is capable of selecting the most qualified candidates irrespective of syntactic error in their resumes. One where a machine understands the variables and events behind why you’ve asked it to carry out a task and is capable of making a right and impressive decision when one, some or all of these variables change because it is capable of deducing cause and effects of events.
Why do I think I can bring this kind of Artificial Intelligence into existence? I got a glimpse of this when my robot Onyx (designed in my undergraduate program) created a “gay bias” after conversations with different male friends, who said negative things about the opposite sex. This caused Onyx to create a model that criticized the female sexuality; preferring the male sexuality to their female counterparts. This incident inspired me to investigate the reasons behind “why words we hear aid formation of our bias”.
Have you ever experienced a toddler not understanding the idea in your message? The thing is a toddler can demonstrate understanding of a message only after passing through series of events that define the message. I’m not a father but I fussed about these series of event when I was training Onyx while at Babcock University. The moment a baby is born, he/she uses his/her senses to take in new information about the world around him/her. Although they may not have the legacy knowledge that adults use to decipher new information, However, he/she is building an archive of knowledge with each passing day.
With improving motor skills, memory gets better, lengthening attention span, improving speech and social skills. They strive to understand the connection between cause and effect, they break even and have a greater understanding of their world. This milestone increases their desire to explore and learn the logic that forms the basis for the things they are witnessing. For instance, when a baby stares at a repeated action, they eventually get bored, but the moment the slightest change in this action occurs, they get amused and curious again.
Hence, they start to look for the logic. As their understanding of cause and effect deepens, they tend to investigate and experiment, anticipating outcome of events. Repetition of this series of event play a big part in how they acquire language and what they understand.
According to some researchers, babies learn to distinguish sound patterns of their own name from other names at 4 ½ months. When a new baby is born we unconsciously bring ourselves to their level. For instance:
- We use concrete names rather than using pronouns.
- Enunciating our words i.e we articulate the various sounds that make up specific words, such as “Ggrraanndd-ppaaa” for “Grandpa, “On-cool” for “Uncle”.
- Repeating words when they are babbling. i.e. When a baby says “tata” while pointing to a ball, we say the word “ball” repeatedly, helping it to stick. Child experts call this “serve and return”. The baby is serving a sound (blab) here and we return it by making the correct sound (word).
- Encouraging them to imitate things we say and how we say them. Making sure we stress on specific words that are key to the event we are describing i.e. “Henry is playing ball” we would stress more on the noun (subject and object) “Henry” and “Ball”.
- Making efforts to understand what they are saying by repeating what you think is being said and asking if you’re right and getting worried when you are unsure you are getting them i.e. When your child cries and mutters a sound and they are stretching their hands out, we get confused if after carrying them they still cry. However, the minute they point at something and you notice their whole body is trying to reach for something, we recognize in that moment what they are saying and we call out the real name and at a point, they just blab the sound you used in describing the object of interest.
Although just a few months old, they tend to grasp the names of objects that interest them rather than whatever the speaker considers important.
According to a psychology research led by Professor Kathy Hirish-Pasek in Temple University, infants were able to learn new words in five minutes with just five presentations for each word and object. These infants paired new words to objects that caught their fancy, regardless of the actual object the speaker was referring to. They automatically assume the word you are speaking goes with the object that they think is interesting and not what you are showing to be interesting.
This is however not surprising as interest is a key driver in learning.
I remember growing up with my cousin whose father is a pastor. They would refer to a Microphone as a “Praise the Lord”, meanwhile, I interpreted it to have a completely different (and erroneous) sound. This situation of contrasting perspectives would result in a mild argument that was eventually resolved by an adult who then explained what the microphone does. Then it all started to make sense.
Sometimes, we fail to notice what our little learners are doing or showing interest in. The fact remains that learning is only interesting when things are meaningful.
All this eventually results in a modeled language between you and your child. Your modeled language may not necessarily have everything right but it is just enough for you to effectively communicate a message to the developing child. This model is called the Motherese. Motherese is a simplified and repetitive type of speech, with exaggerated intonation and rhythm, often used by adults when speaking to babies. Just like other social registers, it is highly structured, and not random and irregular as Noam Chomsky an American linguist and cognitive scientist would have us believe; for there is a set correlation between a fully formed human language and Motherese. If a machine is capable of modeling a Motherese with a patron, then Natural Language Processing (a subset of Artificial Intelligence dedicated to building systems capable of understanding language) has empowered machines to take a moonshot at being more human like.
