Deciphering AI Speak: A Primer for English Language Educators
by Joshua M. Paiz and Ilka Kostka
If you are like us, AI (artificial intelligence)
has come to play some role in a growing number of professional
conversations—either at conferences, in meetings, or during breaks with fellow
teachers. In these conversations, you have likely heard an almost dizzying
array of jargon that might get in the way of clear communication. In this
article, we offer a simple guide to some of the more common words and ideas
surrounding AI so that you can better participate in these conversations and
continue to advocate for yourself and for your multilingual learners. From
there, we offer advice for gaining familiarity with AI tools.
AI Jargon
The field of AI is a subfield of computer sciences
that is focused on creating intelligent computer systems. To begin, we draw
your attention to some fundamental terms that come up in many conversations
about AI and AI-based systems. In Table 1, to help you more comfortably engage
in AI-focused conversations, we list the most common AI-related terms, working
definitions, and, where relevant, examples of AI platforms.
-
Large language model (LLM): An AI system trained
to understand and generate human-like text/speech, aiding in tasks like
translation, summarization, and conversation.
-
Machine learning (ML): A field of AI that enables
computers to learn from data and make predictions or decisions without explicit
programming.
-
Natural language processing (NLP): An
interdisciplinary field at the intersection of AI and machine learning that
focuses on teaching machines to understand and interpret human
language(s).
-
Transformers: AI technology that can consider a
whole text, being aware of relationships between ideas across the entire text
and how this creates unique context for meaning making, resulting in more
accurate and coherent text generation.
-
Generative Pre-trained Transformer (GPT):
Advanced AI model that uses the transformer architecture to generate human-like
text based on a given prompt. It was trained on a set of data containing more
than 400 billion tokens of text and code.
-
Pathways Language Model 2 (PaLM2): Large language
model developed by Google that was trained on more than 4.6 trillion tokens of
text and code and is seen as a more capable model than the most recent model
GPT (i.e., GPT-3).
Table 1. AI Jargon for English Language Teaching
Educators
Getting Started With Generative AI
The use of generative AI in English language
teaching (ELT) is relatively unexplored; yet, as technology continues to
rapidly develop, educators will need to keep up with the latest developments.
Following, we offer suggestions for building AI literacy skills and better
understanding of how they work. Because of its ease of use and because it is
currently freely available, we take ChatGPT as an example here of a path to
skilling up. However, our advice can be generalized to many chat-based AI
tools.
1. Begin With Free
Tools
With the availability of free versions of many AI
programs, teachers should not have to pay for most of these tools. For
instance, ChatGPT
3.5 is currently freely available; users simply need to create an
account and log in with their username and password. If a teacher decides to
use a tool more regularly or extensively, they might opt to pay for a more
powerful version to maximize its benefits through more capable models with
expanded abilities.
2. Experiment and
Explore
ChatGPT is simple to use. Users type a prompt into
a box that resembles a chat box, and ChatGPT immediately produces human-like
responses to the prompt. Considering this, we suggest appreciating the
conversational nature. For instance, you can ask ChatGPT for clarity or you can
continually refine your prompt until the AI produces a sufficient response.
When using ChatGPT, for instance, creating effective prompts is critically
important for getting useful output. This process is referred to as prompt
engineering, and it is developing into its own field of
scholarship and practice.
When teachers input effective prompts, they can use
ChatGPT to create a lesson, generate ideas for in-class activities, and
streamline materials creation. Table 2 shows a prompt that has been revised to
push the AI to produce more meaningful output for an imagined academic writing
class. You are welcome to try it for yourself, but because of the nature of
generative AI, you will receive unique output each time you enter the same
prompt. The Appendix (PDF) includes screenshots of these prompts to show how they
appear in ChatGPT, as well as an explanation generated by ChatGPT about when to
use idioms in academic writing and ideas for creating a handout for
students.
Table 2. Generative AI Prompt: Before
and After Engineering
|
Before Prompt
Engineering |
|
Create a handout
on academic idioms. |
|
After Prompt
Engineering |
|
I’m teaching an
EAP class for first-year college students, and I want to teach my students
about how we can use idiomatic expressions in academic writing. I already have
the class planned and will introduce the fact that we typically avoid idioms in
academic writing, but we do use idiomatic expressions like “shed light on” and
“a double-edged sword”. Can you help me come up with a handout of 15-20 common
idiomatic expressions with definitions and examples? |
3. Upskill, Learn, Connect, and
“Play”
Short of getting a master’s degree in computer
engineering, you can learn more about AI and how it works a number of
ways:
With the increasing amount of literature and
resources about AI in education, finding materials is relatively
easy.
Final Thoughts
The release of large language models like ChatGPT
was certainly a shock for many because of its profound implications for
teaching, learning, and work. We would like to emphasize that AI is never a
replacement for classroom teaching and learning, and it is most certainly not a
replacement for the expertise of the English language teacher. Instead, given
the realities of much ELT labor, we see AI as a potentially powerful tool for
the educator, allowing them to leverage their expertise and save time preparing
for instruction.
Our students are also entering an AI-rich world—one
in which AI literacy will permeate the workplace (McKinsey, 2023) and in which
the risks of AI are unknown. As such, preparing them to critically engage with
AI tools and products will become increasingly important. Perhaps more worrying
is the potential for AI abuses, further creating the need for an educated
citizenry who can critically engage with AI-generated material (Vincent, 2023).
For these reasons, ELT practitioners who engage with AI tools can help equip
their students with the skills needed to safely and effectively navigate a new
world in which AI plays a prominent role.
References
McKinsey. (2023). The economic potential
of generative AI: The next productivity frontier [White Paper].
McKinsey & Company.
Vincent, J. (2023, Apr 25). Republicans respond to
Biden reelection announcement with AI-generated attack ad. The
Verge. https://www.theverge.com/2023/4/25/23697328/biden-reelection-rnc-ai-generated-attack-ad-deepfake
Joshua M.
Paiz, PhD, is a teaching assistant
professor of EAP at George Washington University, Washington, DC, USA, and a
teacher educator at Montgomery College and the Community College of Baltimore
County. He holds a PhD in TESOL/applied linguistics from Purdue University. His
scholarly work focuses on LGBTQ+-inclusive pedagogy and the application of AI
and natural language processing to ELT.
Ilka
Kostka, PhD, is a teaching professor at
Northeastern University in Boston, Massachusetts, USA, where she teaches
English language courses to undergraduate and graduate international students.
Her primary interests include technology-facilitated language learning, flipped
learning, and source-based writing instruction. She is the secretary of
Northern New England TESOL, an affiliate of TESOL International
Association.
Previous Article
;){kind=small}
