NLP for Social Good Symposium at University of Liverpool -- Closing Remarks

Slide 1

Slide 1 text

NLP for Social Good Symposium Danushka Bollegala

Slide 2

Slide 2 text

Thank you! • 6 invited speakers • 160+ pa rt icipants! (online + in-person) • Funding/Suppo rt • Depa rt ment of Computer Science, University of Liverpool • Organisers • Procheta Sen, Tulika Saha 2

Slide 3

Slide 3 text

Why talk about Social Good? 3 Images created by DALL-E (OpenAI) h tt ps://www.vice.com/en/a rt icle/wxdawn/the-ai-that-draws-what-you-type-is-very-racist-shocking-no-one

Slide 4

Slide 4 text

Why talk about Social Good? 4 Images generated by DALL-E (OpenAI)

Slide 5

Slide 5 text

Is AI Socially Biased? 5 Images generated by Stable Di ff usion (Stability.AI) “Janitor” “Asse rt ive Fire fi ghter” h tt ps://techpolicy.press/researchers- fi nd-stable-di ff usion-ampli fi es-stereotypes/

Slide 6

Slide 6 text

Why talk about Social Good? 6

Slide 7

Slide 7 text

Why talk about Social Good? 7

Slide 8

Slide 8 text

I 8

Slide 9

Slide 9 text

No content

Slide 10

Slide 10 text

My take on “existential threats” of AI 10

Slide 11

Slide 11 text

My take on “existential threats” of AI • What I hear… 10

Slide 12

Slide 12 text

My take on “existential threats” of AI • What I hear… • We will loose our jobs… 10

Slide 13

Slide 13 text

My take on “existential threats” of AI • What I hear… • We will loose our jobs… • AI will take over humans… 10

Slide 14

Slide 14 text

My take on “existential threats” of AI • What I hear… • We will loose our jobs… • AI will take over humans… • NLP is over (with ChatGPT) 10

Slide 15

Slide 15 text

My take on “existential threats” of AI • What I hear… • We will loose our jobs… • AI will take over humans… • NLP is over (with ChatGPT) • My PhD research topic is over… 10

Slide 16

Slide 16 text

What we (NLP researchers) should do 11

Slide 17

Slide 17 text

What we (NLP researchers) should do • Be realistic • Look at the real threats that would otherwise get overlooked because of the hyped threats. 11

Slide 18

Slide 18 text

What we (NLP researchers) should do • Be realistic • Look at the real threats that would otherwise get overlooked because of the hyped threats. • Be responsible • If you believe there is a non-zero probability of a risk, think what you can do to address it. 11