AI in education done properly
Download as PPTX, PDF•1 like•511 views
This document discusses several topics related to properly implementing AI in education, including: 1) Ensuring AI teacher evaluation and models are not biased toward specific demographic groups or teaching styles. 2) The importance of data quality when training AI models, such as removing duplicates and standardizing formats. 3) The need for explainable AI models. 4) Examples of non-machine learning AI applications, such as an automated study topic scheduler. 5) A reminder that we have a choice in how AI is designed to have a positive impact.
AI in education done properly
- 1. AI in education done properly Fariz Darari ([email protected]) Faculty of Computer Science, UI Dec 13, 2019 Kemdikbud, Senayan
- 3. Bapak Budi adalah guru yang berdedikasi tinggi. Bapak Budi selalu memberi semangat ke murid-muridnya. Satu hal kecil yang kurang dari beliau adalah latihan soal yang diberikan terkadang agak membingungkan. AI-assisted teacher evaluation by students Dedikasi Memberi semangat Kualitas latihan soal
- 8. Unbiased AI: Demographic groups • Indonesia is a vast country: not only Java island! • Data for AI learning must be kept balanced • In other words, AI should NOT be biased toward a specific demographic group • For example, taking only data from Jakarta schools to build AI models would give some bias when the models are applied to non-Jakarta schools
- 9. Unbiased AI: Different teachers, different styles • Every teacher is different: they have their own, unique styles! • AI should take teaching styles into account • In other words, there must be NO biases toward a specific style
- 10. Data quality matters for training AI models • Garbage in garbage out principle: • Incomplete data • Inaccurate data • Out-of-date data • Get data into right shape: • Remove duplicates, incorrect data • Standardize data formats • Update data, don't let data become expired • Reduce noise • Anonymize data
- 11. Explainable AI
- 12. Explainable AI
- 13. Automated study topic scheduler Given the following study topics and their dependencies, find a proper schedule!
- 14. Automated study topic scheduler Given the following study topics and their dependencies, find a proper schedule!
- 15. Not all problems have to be tackled by "AI" • Well, AI is actually more than machine learning (ML) or deep learning (DL): ML/DL is a branch of AI • The example you just saw is actually AI! • Definition of AI by John McCarthy (Inventor of AI): "The science and engineering of making intelligent machines"
- 17. Our lives at hand of AI, our choice to design good AI. Thanks!
Editor's Notes
- #2: AI done carefully Audience includes Pak Dr. Muktiono Waspodo (Kepala Litbang Dikbud)
- #3: Text mining: AI technology leveraging natural language processing to transform texts into knowledge/insights Evolution should be evaluation https://onlinelibrary.wiley.com/doi/abs/10.1002/widm.1332
- #4: Aspect-based sentiment analysis https://aylien.com/text-api/aspect-based-sentiment-analysis/ https://www.iconfinder.com/iconsets/emoji-22
- #5: https://cacm.acm.org/careers/238947-flawed-algorithms-are-grading-millions-of-students-essays/fulltext
- #6: https://onlinelibrary.wiley.com/doi/full/10.1002/ets2.12192 https://cacm.acm.org/careers/238947-flawed-algorithms-are-grading-millions-of-students-essays/fulltext
- #7: https://cacm.acm.org/careers/238947-flawed-algorithms-are-grading-millions-of-students-essays/fulltext
- #8: More creative + more fluent, effective use of less-sophisticated words When the ETS researchers compared the average difference between expert human graders and E-rater, they found that the machine boosted students from China by an average of 1.3 points on the grading scale and under-scored African Americans by .81 points. Those are just the mean results—for some students, the differences were even more drastic.
- #10: Ada gaya mengajar super disiplin, ada pula gaya mengajar santai, tapi tetap mengena Guru senior (wibawa, pengalaman banyak) vs. guru muda (santai seperti teman, relatable)
- #11: Context: AI learning https://www.fuzzylogx.com.au/fuzzy-friday/fuzzy-friday-part-20/ https://www.forbes.com/sites/cognitiveworld/2019/03/07/the-achilles-heel-of-ai/#1cfb2e8a7be7
- #12: https://www.kdnuggets.com/2019/01/explainable-ai.html
- #13: https://www.kdnuggets.com/2019/01/explainable-ai.html
- #14: https://courses.cs.washington.edu/courses/cse326/03wi/lectures/RaoLect20.pdf
- #15: https://courses.cs.washington.edu/courses/cse326/03wi/lectures/RaoLect20.pdf
- #16: Intelligence is the computational part of the ability to achieve goals in the world. advanced algorithms, logic-based AI http://jmc.stanford.edu/articles/whatisai/whatisai.pdf
- #17: http://aima.cs.berkeley.edu/index.html
- #18: Add little issue: to assist/augment rather than to replace https://insights.dice.com/2019/11/05/ageism-technology-hiring-stop-good/