DePollo 1

Jenna DePollo

Professor Malcolm Campbell

WRDS 1104

10 Nov. 2021

“Hey Siri, Can You Do My Spanish Homework for Me?”: Ways Artificial Intelligence is

Becoming Smarter Than Humans

Have you ever wondered how Siri can talk back to you? Or how she even understands

your questions? Well, many of us, including myself, use artificial intelligence like Siri and Alexa

in our everyday lives. But we never ask ourselves how all of it is even possible. With a simple

command, our devices perform tasks without us having to do anything. The concept has become

very normalized in society. But how does it all work? Siri and other artificial assistants were

created using many forms of technology like speech recognition and natural language processing

to have “human-like” conversations with us.

The idea of conversing with non-human objects or giving them human characteristics is

called Anthropomorphism. You probably do it everyday without even thinking about it. These

virtual assistants have become so human-like in their speaking that they have become better

speakers than some humans. The continuous use of algorithms and technology may eventually

make them smarter than us.

But still the why and how question is not answered… Previously mentioned speech

recognition and natural language processing are the two technologies behind Siri and other AI

(Artificial Intelligence). Speech recognition is the act of converting human speech into text form.
 DePollo 2

Sounds easy enough. Not at all. This process is very strenuous and requires highly skilled

engineers. The way every person speaks and writes is very different. Accents, speed, pitch, all of

these factors have to be taken into consideration. The engineers at Apple need to collect a lot of

data to create their speech recognition models for Siri. Due to continuous advancements in

technology, the speech recognition error rate is only around 10%. This means that Siri can

understand many types of languages and accents.

The other part of Siri development is natural language processing. Natural language

processing is the connection between linguistics and technology. Together they break a barrier

and work as one. Siri is very interesting because through software and coding she is able to retain

information just like humans. In the article Natural Language Processing ft. Siri, Komal Saini, a

student studying Computer Science at the University of Toronto, uses the perfect analogy to

explain how Siri is “trained” to retain information. Saini talks about how a toddler responds to

getting in trouble for biting someone. After a while they start to realize the consequences of their

actions and train themselves not to do it again. This is the same thing for programming Siri, the

software picks up on the repeated actions and learns to be trained to perform certain tasks.

Another way Siri and other AI are able to retain information is through Named Entity

Recognition. Named Entity Recognition, is a process that adds general knowledge and allows the

machine to understand questions and given tasks. This type of recognition makes it easier for

processing text. Obviously just like everything, the software is not perfect and sometimes is

unable to complete tasks or answer questions. Technology is advancing everyday and coming out

with new ways to make our lives easier.

A recorded lecture from Stanford University goes into even more detail about natural

language processing. The lecture was given by Christopher Manning, who studied linguistics and

computer science. The majority of what he was saying revolved around the human language and

why we do and speak the way we do. Manning said that the majority of our language is made up

of three factors. One being sound, two gestures, and three images. The human language seems

easy, but can become very complex to replicate by artificial intelligence. The main idea of virtual

assistants is to understand written or verbal instructions and then respond back with something

helpful. Difficulties that engineers and software developers face is figuring out how to program

“intent.” In languages used across the world there are multiple ways and styles to say something,

that sometimes us humans don’t even understand.

None of what Siri and AI can do would be possible without technology rapidly

developing all the time. At first Siri was only able to speak one language and perform a few

tasks, now she is online shopping, ordering ubers, setting appointments, etc. Engineers focussing

on the human language and how it was constructed, then creating virtual assistants that can

respond right back to us continuously blows my mind. Siri and other AI were very quickly

normalized into society. Giving them power to influence the way we live our lives. A new way it

is being incorporated into our lives is through learning. Learning apps have been created and use

AI to teach people. Most of these learning apps are targeted toward children and the upcoming

generations. Repeated algorithms make it easy for developers to create the software. Marcel

Pikhart, the author of a peer reviewed journal “Intelligent Information Processing for Language

Education: The Use of Artificial Intelligence in Language Learning Apps”, thinks we should be

using AI to teach our children in schools.

I agree with Pikhart. The world continues to surround and grow with technology constantly at

our fingertips. Not only does it make learning easy, but children want to learn on their iPads and

computers. With such a large demographic using technology and AI, it shouldn’t come as a

shock if it was implemented into learning. Pikhart also thinks having language learning apps

would be a helpful tool. Softwares like natural language processing are becoming more and more

common, making it more simple to create them.

Pairing along with natural language processing is deep learning. Deep learning is a type

of machine learning and AI, which imitates how humans gain knowledge. The Stanford

Professor, Christopher Manning said that “Deep learning is a subfield of machine learning.”

There are many reasons why artificial intelligence should use more of this in development. First

off, manually designed software is hard to make efficient and takes a long time to create.

However, deep learning features are easy to change and you can learn it very quickly.

I found that discovering more about what was behind the voice and functionality of Siri

was very fascinating. Multiple types of software and developmental techniques used to create a

human-like feature on technology. This is important because as we grow as a society and

technology develops with us, we need to understand how it's implemented into our learning and

language. This means that we need to learn more about what we use in our everyday lives. In the

future I feel like my topic will be more widely discussed. Soon we may not even have teachers

teaching class, it could be artificial intelligence.

Works Cited

Goel, Aman. “How Does Siri Work? the Science behind Siri.” Magoosh Data Science Blog,

1 Feb. 2018, https://magoosh.com/data-science/siri-work-science-behind-siri/. Accessed

14 Nov. 2021.

Hye Rin Shim, and Byoung Gwan Kim. “The Effect of Customer Value on User Satisfaction

With Dialogue Characteristics of Apple’s Intelligent Agent Siri.” Journal of

Organizational and End User Computing, vol. 32, no. 1, IGI Global, 2020, p. 62–,

doi:10.4018/JOEUC.2020010104.Accessed 24 Oct. 2021.

Massaro, Toni M., and Helen Norton. "SIRI-OUSLY? FREE SPEECH RIGHTS

AND ARTIFICIAL INTELLIGENCE." Northwestern University Law Review, vol. 110,

no. 5, 2016, pp. 1169-1194. ProQuest, Accessed 24 Oct. 2021.

Pikhart, Marcel. “Intelligent Information Processing for Language Education:

The Use of Artificial Intelligence in Language Learning Apps.” Procedia Computer

Science, vol. 176, Elsevier B.V, 2020, pp. 1412–19, doi:10.1016/j.procs.2020.09.151.

Accessed 24 Oct. 2021.

Saini, Komal. “Natural Language Processing Ft. Siri.” Medium, MyTake, 16 Sept. 2019,

https://medium.com/mytake/natural-language-processing-ft-siri-2bc7b854a2a3.

Accessed 14 Nov. 2021.

Stanford University School of Engineering. “Lecture 1 | Natural Language Processing with Deep

Learning.” Youtube. Commentary by Professor Christopher Manning, 3 Apr 2017.

https://youtu.be/OQQ-W_63UgQ. Accessed 14 Nov. 2021.