A sizable number of teachers are skeptical or uncertain about the use of artificial intelligence in schools, according to recent survey results from Pew Research Center.
Of 2,531 U.S. public K-12 teachers polled in fall 2023, 25% said they see more harm than benefit in using AI tools in K-12. Meanwhile, 32% said there is about an equal share of benefit and harm in using the technology in school, and just 6% said they see more good than harm. A remaining 35% said they were unsure.
In a separate Pew survey of 1,453 U.S. teens ages 13 to 17, two-thirds said they know of the generative AI tool ChatGPT. For teens who had heard of ChatGPT, 19% said they have used it to assist them with school assignments.
Older students were more likely to use ChatGPT, with almost a quarter of 11th and 12th graders saying they have done so, compared to 17% of 9th and 10th graders and 12% of 7th and 8th graders.
Among teens who have heard of ChatGPT, 69% said it’s acceptable to use it for researching new topics, 39% said it’s fair to use it for solving math problems, and 20% view it as acceptable for writing essays.
Some 18% to 24% of teens said they were still unsure if it is acceptable to use ChatGPT in these areas.
Teachers’ doubts regarding the benefits of using AI in schools come as separate findings published in April by Rand Corp. and the Center on Reinventing Public Education found just 18% of educators reported using AI tools in their classrooms. And as suspicions of AI-related plagiarism rise, student discipline due to generative AI use has also gone up from 48% to 64% between the 2022-23 and 2023-24 school years, the Center for Democracy & Technology found.
Educator use of AI content detection tools also jumped from 38% to 68% within the past school year, according to CDT.
Since Turnitin, a grading and plagiarism detection platform, launched its own AI content detection feature in April 2023, the company found that 11% of the 200 million student papers reviewed contained at least 20% AI writing, while 3% of papers were found to include at least 80% AI writing.
Even so, CDT has flagged research finding that similar kinds of AI detection software can be inaccurate when differentiating between AI-generated and human-written content.
