�from AI Literacy to �AI Agency��Teemu Roos�University of Helsinki

26.3.2024

privacy

social media

AI

algo-rithms

trust

filter

literacy

machine learning

skills

data

cyber-security

agency

profile

Teemu Roos – who?

Professor, Department of Computer Science,�University of Helsinki (UH)

Leader of the Education Program, Finnish Center for AI (FCAI)

PhD (UH) 2007, visiting scholar MIT, Berkeley, Cambridge, ...

research:

• statistical machine learning, applications in neuroscience, astrophysics, epidemiology, ...

teaching:

• 15 years teaching AI courses
• Lead Instructor of the Elements of AI MOOC, 1,2M+ signups
• co-PI in the Generation AI project 2022-2028, ca. 5,6M€

GENERATION AI

• The Generation AI project tackles three interwoven societal challenges:
• weakening feeling of security, control, and free will in a datafied society
• eroding trust on public authorities, media, science, civil society, and fellow citizens
• rapidly growing inequalities related to ICT ability and agency.

​

• Strategic Research Council: 3+3 years, total 5.4 M€�
• Consortium PI: Prof Matti Tedre, University of Eastern Finland

​

• Web pages: https://www.generation-ai-stn.fi

​

​

​

Computational Thinking 2.0 (Tedre, Denning, Toivonen et al.)

Google Generation AI Teachable Machine

https://tm.generation-ai-stn.fi/

Google Generation AI Teachable Machine

https://tm.generation-ai-stn.fi/

data

training

classifier

trial & error

bias

Basic AI Literacy

1. What concepts or "thinking tools" are needed to understand how AI systems work�������
2. What activities / teaching practices could illustrate the dynamics of social media?
• virality, polarization, misinformation

data

training

trial & error

filter

classifier

bias

polarization

misinformation

Summing up so far

• New thinking tools needed for the ”Generation AI”
• Likewise, new concepts needed in (CS) education:
• Computation Thinking 2.0
• …
• Generation AI Project (STN):
• Materials ⬄ Teaching methods ⬄ Research

��Access < Literacy < Agency�(Matti Tedre, personal communication, 2023)

Photo by Teemu Roos

Access: One Laptop per Child (OLPC)

• Founded 2005: Negroponte, Papert, Jepsen, Kay, …�
• “Expecting a laptop to cause such a revolutionary change showed a degree of naivete, even for an organization with the best of intentions and the smartest of people”

Critical AI Literacy

• Many courses focus on things like:
• “basic understanding of AI”
• “AI resources, tools and technologies”
• “hands-on AI-related activities”
• “gain strategic advantage through [AI]”�
• Although many also mention ethical�aspects, few actively encourage the�user towards “critical AI literacy”��(Maha Bali, “What I Mean When I Say Critical AI Literacy”, 2023)

Ng, Leung, Chu & Qiao, “Conceptualizing AI literacy: An exploratory review”, Computers and Education: Artificial Intelligence, 2021

# articles per year on AI literacy

Generation AI Social Media Machine

filter

polarization

misinformation

13

active

passive

positive

negative

–5

+5

+5

–5

We shouldn’t necessarily try to move people towards a more positive or negative attitude on AI.

However, we should try to move people towards a more active position on AI.

Data Agency

The Guardian, 14.8.2020

• Free online course, about 30 hours
• Joint project by the University of Helsinki and�MinnaLearn since 2018

​

• 1 278 151 registered users� 158 543 completed�
• over 170 countries, 37-56% women,�21% over 45 yrs�
• follow up course Building AI
• a new update on LLMs (9/2023)

15

26.3.2024

Elements of AI www.elementsofai.com

​

Summary

Literacy is necessary but not sufficient. �=> Agency:

• Ethical and regulatory decisions are inherently political questions.
• We (educators/researchers) can’t make such decisions on other people’s behalf.
• Therefore, we should encourage people make their own, informed decisions and to be active.

Photo by Kadir Altıntaş on Pexels

Image: Clarote & AI4Media / Better Images of AI / Power/Profit / CC-BY 4.0