tilburg_ai.qmd

---
title: "Tilburg University:<br> AI-ready?"
subtitle: "Educate, Encourage, Experiment, Invest"

format:
  revealjs: 
    logo: img/logo.png
    slide-number: true
    slide-level: 3
    controls: true
    transition: fade
    background-transition: fade
    preview-links: auto
    theme: theme.scss
    #incremental: true
    title-slide-attributes:
      data-background-image: "img/title-background.jpg" 
      data-background-opacity: "0.15"
      data-background-size: cover
      data-background-color: "#036"
---

## A Vision of the Near Future

![<font color=grey>"Can you create a picture of a flourishing campus at Tilburg University? Add students and a futuristic skyscraper," created by [OpenAI's DALL·E](https://labs.openai.com).</font>](img/DALLE_flourishingcampus_many_students_really_futuristic_buildings2.png)


## A Vision of the Near Future

- AI-facilitated blended learning
- Algorithmic allocation of students to courses
- Personalized and interdisciplinary programs

::: goal
AI technology developed by students and faculty has revolutionized higher education.
:::

# Agenda

1.  What is AI?
2.  How do we already use AI?
3.  What are new ways in which we can use it?
4.  How to get AI-ready?


## Who am I? {.smaller}
- [Hannes Datta](https://hannesdatta.com); Associate Prof. at TiU since 2013
- Working in marketing research, studying digital markets
- Passion for innovation (e.g., TestVision, Blended Learning, [Tilburg Young Academy](https://www.tilburguniversity.edu/nl/onderzoek/tilburg-young-academy), [Tilburg Science Hub](https://tilburgsciencehub.com))

. . .

**Disclaimer**

- I'm not on an "AI mission" - I got in this accidentally 
- Not an AI researcher (but: I love technology!)
- Researcher from TiSEM (but: involved 10+ faculty and staff members, across schools and divisions)

<!--- Focus on AI &rarr; University (not: AI &rarr; workplaces)-->

# 1. What's AI?{background-color="#036"}

## AI is everywhere

:::: {.columns}

::: {.column width="50%"}
__In daily life__

- Search and recommendation (e.g. Bol.com)
- Face recognition (e.g., iPhone)
- Navigation (e.g., Google Maps)
- Voice assistants (e.g., Alexa)

:::

::: {.column width="50%" .fragment}

__At Tilburg University__

- Plagiarism detection (e.g., TurnItIn)
- Text editing (e.g., "Editor" in Word or Grammarly)
- Automatic messages (e.g., Teams)
- Voice assistants (e.g., Outlook's "play my email")
:::

::::

::: {.notes}
- AI is not as new as we think
- Algorithms are everywhere
- Some examples of AI applications that you may have already come across... many more at futurepedia.io
:::


### And then came ChatGPT... {.smaller}
**What's new?**

- __General-purpose AI__ (vs. specialized AI)
- __Language__ model (*seems* to understand)
- Trained on __massive data__ (with internet connection at Bing)
- __Conversational__ (iterative input, "creative" output)

. . .

**How to use it?**

- Open [chat.openai.com]() 
- Login & refresh until you're in
- Start a conversation!

::: {.notes}
  - Prompting (i.e., encourage to __act__ like someone or something)
  - Asking (i.e., your specific task)
  - Refinement
:::

### 

![<font color=grey> Examplary use of ChatGPT (for preparing my exam): Prompt, ask & refine </font>](img/chatgpt-mcquestions.png)

### 

![<font color=grey> Examplary use of ChatGPT (for preparing my exam): Prompt, ask & refine </font>](img/chatgpt-mcquestions2.png)


# 2. How is AI currently used at Tilburg University?{background-color="#036"}


## How we use AI in teaching {.smaller}

- Course design and management
  - Creating and revising exam questions
  - Generating and/or improving syllabi and lecture outlines
  
- Content delivery
  - Finding examples and case studies
  - Creating and improving slides
  
- Automation
  - Connect different systems (e.g., Canvas API)
  - Semi-automated grading


::: goal
We leverage AI to efficiently improve education through engaging, multimodal, and high-quality content.
:::

### 

![<font color=grey>Lecturing using AI: text-to-speech + avatar animation ([view on YouTube](https://youtu.be/fL6QlPaXY_4), by Anna Paley, TiSEM)</font>](img/aivideo.png)


## How our students use AI {.smaller}

- Searching information (e.g., much faster than Google)
- Note-taking during class (e.g., voice-to-text; typing & filling in gaps)
- Drafting, fixing or annotating computer code (e.g., "I have a bug in this code - do you spot it?")
- Ideation and problem solving (e.g., "which `R` package to use for X?")
- A "virtual" teacher, 24/7, immediate response (e.g., "why is X important?", "can you explain how X works?", "I want to learn more about X. Where to start?")
- Editor (e.g., finding good titles & writing abstracts, feedback on writing style, developing paragraph structures, synthesizing; CV writing)
- Prompt engineering (e.g., more specific goals, more context, assign role and persona, feed with information, "can you confirm?")

::: notes

 (e.g. Hey chat GPT, what do you think of this letter/email in relation to its goal/objective?)
 
:::

### 

![<font color=grey>Text editor popular among students (Quillbot)</font>](img/quillbot.png)


## How we use AI in research {.smaller}

- As a subject of research (e.g., method development, societal impact, legal landscape)
- Editing, proofreading (e.g., cutting words)
- Knowledge discovery and ideation (e.g., background research, key references, outline generation)
- Automation (e.g., filling in forms, answering emails)
- Research assistant (e.g., generating stimuli, summarizing articles, transcribing interviews)
- Writing computer code (e.g., using GitHub copilot)

### 

![<font color=grey>Using GitHub Copilot for suggesting "the next line of code"</font>](img/copilot.png)

# 3. Novel applications of AI{background-color="#036"}

## The "next big things" {.smaller}

**Release of GPT-4**

- latest version, ["GPT-4"](https://openai.com/product/gpt-4), was recently released
- major leap in AI: creativity, advanced reasoning
- passing percentiles: bar exam (90th), GRE quant. (80th), GRE verbal (99th)

. . .


**Integration and automation using Application Programming Interfaces (APIs)**

- API ecosystems ("linching pin" of the internet economy; e.g., Facebook, Spotify, ...)
- Entire functionality of ChatGPT (and more) can be programmatically accessed
- GPT-4 already used in education: student feedback (Khan Academy), conversations in foreign languages (Duolingo)

. . .

::: goal
&rarr; AIs become smarter and can be integrated, fine-tuned & automated.
:::

## Future use cases {.smaller}

- Low-hanging fruits
  - Improving assessment (e.g., more formative testing, generation of questions)
  - Aligning course content with "blended" learning strategy (e.g., updating syllabus, introducing more interactive moments)
<br><br>
- Challenging but possible with today's technology
  - Chatbots (e.g., student help desk, IT help desk, but also study & career coach)
  - Automating assessment (e.g., written feedback, automated grading)
  - "Recycling" existing course content (old lectures) by creative cutting or format conversions

<!-- Example here on automated improving of exam questions 

(TestVision &rarr; select "bad" exam questions &raar; improve them with a "good" prompt &raar; create new draft questions in TestVision &rarr; send email to teacher, informing about update and asking for review)

-->

## Risks and opportunities {.smaller}

:::: {.columns}

::: {.column width="50%"}

__Risks of embracing AI__

- ChatGPT = copying and cheating?  <!--(but: on-campus exams, "citability" of ChatGPT)-->
- "Will students still learn X?" <!--(but: more feedback, more complicated problems)-->
- How to adapt teaching methods and assessment?<!--(but: also Powerpoint and Covid changed teaching)-->
- Quality checks when scaling up?<!-- (but: firms can do it, already!)-->
- Less physical interaction?
- Impact on staffing? <!--(but: we can redefine roles) <!-- , focus on new content and student coaching-->

<!--- Lack of trustworthiness (but: important 21st century skill to detect it!)
-->
:::

::: {.column width="50%" .fragment}

__Opportunities of embracing AI__

- Using AI is relevant for our students' future and society at large
- Learning how to "use" an AI is not self-evident: that's education, too!
- Offer novel ways to deepen knowledge and foster curiosity
- Allow researchers to stay productive
- Aligns with and promotes our strategy (e.g., blended learning, taking courage )

:::

::::

. . .

::: goal
Initial investment of time, effort, and money. But can enhance quality of education and research.
:::

<!--
- Laziness of employees
- Less interaction in class
::: goal
- Investment in improving education and research
:::
-->

# 4. Making Tilburg<br>AI-ready!{background-color="#036"}

## Making Tilburg AI-ready! {.smaller}

- (1) Educate staff and faculty
  - basic use of ChatGPT & use cases
  - automation with no-coding (e.g., Power Automate) or coding tools (e.g., on Datacamp)
<br><br>
- (2) Encourage use of AI
  - campus license for ChatGPT --- promote use among staff, faculty and students
  - build a community, spanning schools and stakeholder groups, to exchange experiences

## Making Tilburg AI-ready! {.smaller}

- (3) Experiment what works (and what does not)
  - gather (and publish) empirical evidence
  - open up programmatic access to TiU infrastructure (e.g., app development)
  - automate successful prototypes (e.g., course overhaul, assessment)
<br><br>
- (4) Invest in "moonshot" projects
  - Personalization of admission requirements for courses and programs
  - "Algorithmic" and interdisciplinary study programs + blended learning
  - Real-time scheduling (including students' study rooms)
  - "Smart" budget and teaching allocation algorithms


## Conclusion {.smaller}
- Students and some faculty already explore the use of AI in education and research
- AI offers numerous benefits, but also risks that need to be addressed
- To fully leverage AI, we need a clear "AI strategy" (with goals and budget)

<br>

::: goal
Let's embrace AI with __courage__ and __curiosity__ to increase the quality of our education and research.
:::

# Thanks!

![QR Code](img/qr.png){width=150px}
<br>
Download slides at [https://tilburg.ai]().

:::{.columns}


::::{.column width=70%}

::: {.textbox style="padding: 0px; font-size: 14px;"}
I __acknowledge the constructive input__ from Ellen Dreezens (TSB), Phillip Paiement (TLS), Tobias Klein (TiSEM), Anna Paley (TiSEM), Samuel Stäbler (TiSEM), Bob van den Brandt (TiSEM), Anton ten Klooster (TST), Bas van Nunen (IT), Puck Wildschut (LIS/DCC), Niels Boonstra (Library Science Skills), and Diego Sanchez Perez (Tilburg Science Hub).

I have used DALL·E for image generation and exemplary screenshots are from ChatGPT.

I thank Paulina Ambroziak (Tilburg Science Hub) for her research assistance.
:::


::::

::::{.column width=10%}

::::
::::{.column width=20%}


::::
:::