Exploring the Future of Education with Multimodal AI: Insights from a Study on Generative AI and Learning

As we explore the future of education with multimodal AI, let's reflect on how this technology can transform learning experiences for students of all ages.

1. Steph is going to talk about the future of Education with multimodal AI at the AI Engineering Summit.
2. Early literacy rates need improvement, with only 70% of 10-year-olds worldwide able to read and understand a simple story.
3. A significant learning gap was observed during Covid, with 60% of children and teenagers left behind in math and reading.
4. The issue is not limited to K-12; adults also need reskilling.
5. Multimodal AI has the potential to transform education, as students are already using these tools for homework and prefer them over human tutors.
6. The first AI generation has been growing up with AI since 2015, influencing their perception of AI devices and their intelligence.
7. Steph researched youth perception of voice assistants, chatbots, and smart toys at MIT in 2015, showing that youth perceive AI devices as friendly but have different perceptions of their intelligen
8. A child's perception of AI influences the type of queries they ask and how much they trust the answers.
9. Cultivating AI literacy and a critical understanding of this technology is essential for students to understand how it works, its limitations, and how to use it responsibly.
10. Steph built an open-source platform called Cognates in 2016, expanding Scratch (the largest platform for coding for kids), allowing children to program smart lights, voice assistants, train custom
11. A longitudinal study in public and private schools showed that AI learning activities increased critical understanding of AI, making students more skeptical of AI's "smarts" after learning how it
12. Tech-savvy parents recognize the importance of teaching their children 21st-century skills, including AI literacy, as technology will play a significant role in their lives.
13. Engaging students in the scientific process and allowing them to form hypotheses and test them is crucial for fostering AI literacy.
14. Parents need to learn alongside their children, and the pandemic provided an opportunity for families to learn together when kids were stuck at home.
15. Co-pilot programming for families can help generate coding ideas with an AI friend, express and elaborate ideas in code, support a creative coding identity, encourage collaboration between parents
16. Evaluating generative AI models to design a co-pilot for programming for families can identify what works and what doesn't based on user preferences.
17. Scratch's top generative AI models are good at generating explanations, giving ideas or questions to help explore games, and creating benchmarks for measuring these capabilities.
18. Multimodal AI can become a creative sidekick in Minecraft, games, physics simulations, science simulations, and other applications, providing ideas based on user input and helping with problem-sol
19. Generative AI models can also serve as learning companions and coaches for various subjects, including math.
20. A Benchmark for math misconceptions was created to evaluate the effectiveness of top generative AI models in identifying common math problems in K-12.
21. AI engineering and AI tinkering should be accessible for all ages, encouraging learning through play and doing.
22. Steph's experiments with the latest Gemini API showcase interactive demos for science, water and CO2 reactions, Earth being hit by an object, math problems, and a curious question-answering system
23. Encouraging curiosity is essential in AI-assisted learning systems.
24. Steph's research is available on her website, and she welcomes questions from the audience.

Source: AI Engineer via YouTube

❓ What do you think? What are your thoughts on the ideas shared in this video? Feel free to share your thoughts in the comments!