Generative AI Can Harm Learning. Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. July, 2024. ![Generative AI Can Harm Learning [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Generative artificial intelligence (AI) is poised to revolutionize how humans work, and has already demonstrated promise in significantly improving human productivity. However, a key remaining question is how generative AI affects learning, namely, how humans acquire new skills as they perform tasks. This kind of skill learning is critical to long-term productivity gains, especially in domains where generative AI is fallible and human experts must check its outputs. We study the impact of generative AI, specifically OpenAI's GPT-4, on human learning in the context of math classes at a high school. In a field experiment involving nearly a thousand students, we have deployed and evaluated two GPT based tutors, one that mimics a standard ChatGPT interface (called GPT Base) and one with prompts designed to safeguard learning (called GPT Tutor). These tutors comprise about 15% of the curriculum in each of three grades. Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48% improvement for GPT Base and 127% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a "crutch" during practice problem sessions, and when successful, perform worse on their own. Thus, to maintain long-term productivity, we must be cautious when deploying generative AI to ensure humans continue to learn critical skills. * HB, OB, and AS contributed equally
@misc{bastani_generative_2024,
address = {Rochester, NY},
type = {{SSRN} {Scholarly} {Paper}},
title = {Generative {AI} {Can} {Harm} {Learning}},
url = {https://papers.ssrn.com/abstract=4895486},
doi = {10.2139/ssrn.4895486},
abstract = {Generative artificial intelligence (AI) is poised to revolutionize how humans work, and has already demonstrated promise in significantly improving human productivity. However, a key remaining question is how generative AI affects learning, namely, how humans acquire new skills as they perform tasks. This kind of skill learning is critical to long-term productivity gains, especially in domains where generative AI is fallible and human experts must check its outputs. We study the impact of generative AI, specifically OpenAI's GPT-4, on human learning in the context of math classes at a high school. In a field experiment involving nearly a thousand students, we have deployed and evaluated two GPT based tutors, one that mimics a standard ChatGPT interface (called GPT Base) and one with prompts designed to safeguard learning (called GPT Tutor). These tutors comprise about 15\% of the curriculum in each of three grades. Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48\% improvement for GPT Base and 127\% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17\% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a "crutch" during practice problem sessions, and when successful, perform worse on their own. Thus, to maintain long-term productivity, we must be cautious when deploying generative AI to ensure humans continue to learn critical skills. * HB, OB, and AS contributed equally},
language = {en},
urldate = {2024-09-13},
author = {Bastani, Hamsa and Bastani, Osbert and Sungu, Alp and Ge, Haosen and Kabakcı, Özge and Mariman, Rei},
month = jul,
year = {2024},
keywords = {Education, Generative AI, Human Capital Development, Human-AI Collaboration, Large Language Models},
}
Downloads: 0
{"_id":"HLiuLAp6HXzezvdQu","bibbaseid":"bastani-bastani-sungu-ge-kabakc-mariman-generativeaicanharmlearning-2024","author_short":["Bastani, H.","Bastani, O.","Sungu, A.","Ge, H.","Kabakcı, Ö.","Mariman, R."],"bibdata":{"bibtype":"misc","type":"SSRN Scholarly Paper","address":"Rochester, NY","title":"Generative AI Can Harm Learning","url":"https://papers.ssrn.com/abstract=4895486","doi":"10.2139/ssrn.4895486","abstract":"Generative artificial intelligence (AI) is poised to revolutionize how humans work, and has already demonstrated promise in significantly improving human productivity. However, a key remaining question is how generative AI affects learning, namely, how humans acquire new skills as they perform tasks. This kind of skill learning is critical to long-term productivity gains, especially in domains where generative AI is fallible and human experts must check its outputs. We study the impact of generative AI, specifically OpenAI's GPT-4, on human learning in the context of math classes at a high school. In a field experiment involving nearly a thousand students, we have deployed and evaluated two GPT based tutors, one that mimics a standard ChatGPT interface (called GPT Base) and one with prompts designed to safeguard learning (called GPT Tutor). These tutors comprise about 15% of the curriculum in each of three grades. Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48% improvement for GPT Base and 127% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a \"crutch\" during practice problem sessions, and when successful, perform worse on their own. Thus, to maintain long-term productivity, we must be cautious when deploying generative AI to ensure humans continue to learn critical skills. * HB, OB, and AS contributed equally","language":"en","urldate":"2024-09-13","author":[{"propositions":[],"lastnames":["Bastani"],"firstnames":["Hamsa"],"suffixes":[]},{"propositions":[],"lastnames":["Bastani"],"firstnames":["Osbert"],"suffixes":[]},{"propositions":[],"lastnames":["Sungu"],"firstnames":["Alp"],"suffixes":[]},{"propositions":[],"lastnames":["Ge"],"firstnames":["Haosen"],"suffixes":[]},{"propositions":[],"lastnames":["Kabakcı"],"firstnames":["Özge"],"suffixes":[]},{"propositions":[],"lastnames":["Mariman"],"firstnames":["Rei"],"suffixes":[]}],"month":"July","year":"2024","keywords":"Education, Generative AI, Human Capital Development, Human-AI Collaboration, Large Language Models","bibtex":"@misc{bastani_generative_2024,\n\taddress = {Rochester, NY},\n\ttype = {{SSRN} {Scholarly} {Paper}},\n\ttitle = {Generative {AI} {Can} {Harm} {Learning}},\n\turl = {https://papers.ssrn.com/abstract=4895486},\n\tdoi = {10.2139/ssrn.4895486},\n\tabstract = {Generative artificial intelligence (AI) is poised to revolutionize how humans work, and has already demonstrated promise in significantly improving human productivity. However, a key remaining question is how generative AI affects learning, namely, how humans acquire new skills as they perform tasks. This kind of skill learning is critical to long-term productivity gains, especially in domains where generative AI is fallible and human experts must check its outputs. We study the impact of generative AI, specifically OpenAI's GPT-4, on human learning in the context of math classes at a high school. In a field experiment involving nearly a thousand students, we have deployed and evaluated two GPT based tutors, one that mimics a standard ChatGPT interface (called GPT Base) and one with prompts designed to safeguard learning (called GPT Tutor). These tutors comprise about 15\\% of the curriculum in each of three grades. Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48\\% improvement for GPT Base and 127\\% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17\\% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a \"crutch\" during practice problem sessions, and when successful, perform worse on their own. Thus, to maintain long-term productivity, we must be cautious when deploying generative AI to ensure humans continue to learn critical skills. * HB, OB, and AS contributed equally},\n\tlanguage = {en},\n\turldate = {2024-09-13},\n\tauthor = {Bastani, Hamsa and Bastani, Osbert and Sungu, Alp and Ge, Haosen and Kabakcı, Özge and Mariman, Rei},\n\tmonth = jul,\n\tyear = {2024},\n\tkeywords = {Education, Generative AI, Human Capital Development, Human-AI Collaboration, Large Language Models},\n}\n\n\n\n\n\n\n\n\n\n\n\n","author_short":["Bastani, H.","Bastani, O.","Sungu, A.","Ge, H.","Kabakcı, Ö.","Mariman, R."],"key":"bastani_generative_2024","id":"bastani_generative_2024","bibbaseid":"bastani-bastani-sungu-ge-kabakc-mariman-generativeaicanharmlearning-2024","role":"author","urls":{"Paper":"https://papers.ssrn.com/abstract=4895486"},"keyword":["Education","Generative AI","Human Capital Development","Human-AI Collaboration","Large Language Models"],"metadata":{"authorlinks":{}}},"bibtype":"misc","biburl":"https://bibbase.org/zotero-group/bethhoppe/5609059","dataSources":["tnM59J55CyNeZmQ8S"],"keywords":["education","generative ai","human capital development","human-ai collaboration","large language models"],"search_terms":["generative","harm","learning","bastani","bastani","sungu","ge","kabakcı","mariman"],"title":"Generative AI Can Harm Learning","year":2024}