1 of 56

Insights on Artificial Intelligence and the Future of Teaching and Learning

2 of 56

Presenters

Pati Ruiz

Senior Research Scientist Learning Sciences and Emerging Technologies

Digital Promise

Kip Glazer

Principal

Mountain View High School, CA

Judi Fusco

Director of Emerging Technologies and Learning Sciences

Digital Promise

Sherry Loftin

Associate Director of Learning Delivery

Digital Promise

3 of 56

Shared Understandings

BE PRESENT.

BE MINDFUL.

BE KIND.

https://bit.ly/aiquestions

Use parking lot for

comments and questions.

4 of 56

To inform and empower education thought leaders on the future of AI technologies and their potential to transform teaching, learning, and assessment.

Objective

5 of 56

Agenda

Introduction & Framing: Artificial Intelligence (AI) in Education
Insights: Keeping Humans Top Priority

ChatGPT, Generative AI, and Rethinking Teaching and Learning for the AI Age
Q&A and Closing

6 of 56

Influence of Artificial Intelligence on society and education

=

Impact of Industrial Revolution

7 of 56

AI in Education:

Let’s talk about the future

8 of 56

Tim Urban predicted this trajectory back in 2015

9 of 56

Tim Urban predicted this trajectory back in 2015

We are here

10 of 56

Tim Urban predicted this trajectory back in 2015

But it feels like we are here

11 of 56

Actual Data shows AI is expanding rapidly

(Peer-reviewed AI Publications (% of Total), 2000-2019)

12 of 56

What is

Artificial Intelligence?

13 of 56

Artificial Intelligence is…

“the theory and development of computer systems that are able to perform tasks normally requiring human intelligence”

“computing that acts independently towards a goal based on inferences from theory or patterns in data”

“automation based on associations”

Judi

First is AI is a branch of CS ..

From studying and developing we get…

This last part tells us there are two kinds of AI, one that is created because we have an understanding of a situation and a second type that comes from data ..

It’s the last one that we herre so much about. Determining what to do based on patterns in the data is how machine learning systems work.

Machine learning systems look at lots of data and try to determine patterns. We’ll come back to this idea of automation based on associations.

____

ChatGPT

LLM knows how words are associated with other words in the context of other words. Different AI model architectures bring in new ways for ChatGPT to make it understand more complex associations, and try to understand the context of the question, but at the heart of AI systems are probabilities and likelihoods. We’ll come back to this in a minute. But first,

14 of 56

AI is so many things

Glossary of AI Terms for Educators

Natural

Language

Processing

X

15 of 56

AI does so many things

Introduction to Artificial Intelligence (AI): A Deep Dive into Machine Learning & Deep Learning | by BangBit Technologies | Medium

16 of 56

Risk from Automation based on Associations

encode and reproduce bias

are not representative

have false correlations

reinforce existing biases

scale up a flawed strategy

have negative consequences

and many more

If Data

They can

Judi

Earlier I described one definition of AI as being automation from association,

Read slide l to r

After scale up → discuss doing things faster and

Later We’ll talk a little more about ChatGPT and some risks it has as it adds some “randomization” to it’s automation based on associations”

If data have false correlations or correlations that exist for some groups but not others, (for example high achieving students sometimes do one type of thing and students who are still acquiring expertise and ability may do other things that the AI system doesn’t support as well. We want to ensure that systems don’t create larger gaps.

---------

One of the things not in the risk slide, but could be is where do we get the data? To be representative, where does it come from? Are the companies using the data compensating the people who give the data?

Data is sacred and precious and AI systems are sometimes given data that they shouldn’t have. Also, there isn’t enough data to train systems in ways that fully represent so who is not being included?

17 of 56

Leading Scholars

Safiya Noble

Marcelo Worsley

Deborah Raji

Ryan Baker

Meredith Broussard

Ruha Benjamin

James Lester

Joy Buolamwini

Timnit Gebru

S. Craig Watkins

Margaret Mitchell

Sasha Costanza-Chock

Cathy O’Neil

18 of 56

Risks in Artificial Intelligence

Risks arise because of how AI is built, who participates in design and testing, lack of disclosure, how it is evaluated, “wild west” free of regulation, and more…

19 of 56

Critical Insights About AI Revealed: Panel Recap

20 of 56

tech.ed.gov/ai/

Nationwide Listening Tours about AI in Teaching | Learning | Assessment | Research

21 of 56

tech.ed.gov/ai/

20

panelists

20+

themes

~ 500

participants

Countless

insights

22 of 56

Key Themes for AI and the Future of Teaching

23 of 56

Benefits

↑

Personalized learning

↑

Support

for neurodivergent students

↓

Administrative tasks

↓

Cognitive load on teachers

24 of 56

Risks

↑

Privacy concerns

↑

Ethical issues

↓

Data security

↑

Reinforcement of biases

25 of 56

Surveillance

Fight against automating our worst impulses

26 of 56

Surveillance

Personalization

&

Balancing tensions between

Freepik.com

27 of 56

Recommendations

Include a diverse range of voices in AI design

Prioritize dignity

Support teachers' decision-making processes with AI rather than replacing them

28 of 56

AI in Education: Prioritize Human-Centered AI

29 of 56

Our schools are, in a sense, factories in which the raw products (children) are to be shaped and fashioned into products to meet the various demands of life.

-Ellwood Cubberley

1st Dean of Stanford Graduate School of Education

1916 book, Public School Administration

Photo: by Stanford School of Education

30 of 56

Leading education influencers thought of children in schools as products.

31 of 56

The Industrial Revolution had collateral damage and children were the losers in many cases.

Detroit Publishing Co, C. C. & Detroit Publishing Co, P. (ca. 1900) Breaker boys,

Woodward Coal Mines, Kingston, Pa. Pennsylvania Luzerne County Kingston United States, ca. 1900. [Photograph] Retrieved from the Library of Congress, https://www.loc.gov/item/2016801353/.

32 of 56

This is our opportunity to get it right.

33 of 56

Intelligence Augmentation

34 of 56

Intelligence Augmentation: technology and humans working together to solve complex problems

35 of 56

AI or Intelligence Augmentation for Education? | Communications of the ACM

Use computation as an additional resource to a human’s abilities and strengths.

Judi

In schools, I really don’t want to replace humans with computers.

I would like it if we could give some of the work that a teacher has to do to a machine, but I would like us to take the time to design it right

so that teachers can say what it si that they want help with and what they the human need to do.

Figuring out what a human should do, what a machine should do and how to work together to solve the complex problems of school will take time.

Machine systems will need to be designed flexibly

I’m mentioning design and the reason I bring it in here is that I think it is essential that everyone of us in the room and the teachers who will use the systems in the classroom

Are included in the design of our new systems for the future of teaching and learning. It’s happened some, but we need to ensure that it happens more.

https://cacm.acm.org/blogs/blog-cacm/251188-ai-or-intelligence-augmentation-for-education/fulltext

36 of 56

ChatGPT, Generative AI, and Rethinking Education for the AI Age

37 of 56

Whole Group Discussion

Share out by raising your hand or typing in the Padlet.

What did you learn and what new wonderings do you have?

https://bit.ly/aiquestions

38 of 56

39 of 56

Fears

Surface Fears vs. Deeper Fears

Surface Fears

Deeper Fears

Cheating / Plagiarism

Losing ability to remember things like phone numbers

Losing ability to concentrate

Survival - How will we earn a living if AI takes over our competitive edge to create value?

Value - What are we good for now if not to produce?

Existential - What makes us special in relation to robots if they can think like humans?

Attribution: Freepik

40 of 56

Ask ChatGPT!

41 of 56

Human-Generated Content

AI-Generated Content

Human and AI Collaboration Zone

42 of 56

Emerging Technology Adoption Framework: For PK-12 Education

circls.org/adoption-framework

43 of 56

Thank you.

44 of 56

We call for our AI conversations to be geared toward helping humans discover the highest expression of themselves instead of replacing them. them.

45 of 56

Augment Human Intelligence - Don’t Replace

46 of 56

Douglas Engelbart believed technology could actually make people smarter.

SRI International, CC BY-SA 3.0 <https://creativecommons.org/licenses/by-sa/3.0>, via Wikimedia Commons

47 of 56

48 of 56

Urgent and Important Conversations

49 of 56

Academic Integrity

50 of 56

Universities are Talking

51 of 56

How do we assess authentic learning?

Photo by Allison Shelley for EDUimages

52 of 56

Resources

53 of 56

Reading List

54 of 56

55 of 56

AI is so many things

Glossary of AI Terms for Educators

Natural

Language

Processing

56 of 56

AI does so many things

Introduction to Artificial Intelligence (AI): A Deep Dive into Machine Learning & Deep Learning | by BangBit Technologies | Medium