K-12 educators are embracing AI to improve assessments. Are you?
by
Did you know that nearly three-quarters of K-12 leaders and educators say they already use artificial intelligence (AI) or plan to deploy it soon? A new national survey finds this information, among other data, a surprise for many observers of K-12 education.
So much of the debate around AI in education has focused on concerns about its use. More than a year after generative AI tools became publicly available, most teachers had yet to use AI in the classroom.
However, most K-12 leaders and educators are embracing AI-powered tools in one area, at least: classroom assessments.
This finding prompts two questions:
- Why are educators turning to AI for assessments when they’ve hesitated otherwise?
- Can AI deliver on educators’ expectations?
Explaining the turn to AI
Let’s start with the first question. The number-one problem the Pearson survey respondents identified is that assessments take too much time. First, assessments eat into classroom instruction time. Outside of class, teachers must choose appropriate assessments, write questions, tabulate results, and analyze the data to determine how to adjust instruction.
The survey shows that educators see AI as a way to reduce the time that some assessments take.
The four most valuable AI features educators identify in the survey correspond to these stages. Respondents believe AI can help by generating questions, scoring open-ended responses, reducing the time spent studying data, and making instructional suggestions based on the results.
Educators see the most value in using AI to reduce time analyzing data. To understand why, look at this other survey finding: only 42% strongly agree that their current assessment data is easy to understand.
Despite such concerns, educators still see the value in assessments. They just want to ease the assessment burden and see AI as a way to do so.
Can AI fulfill expectations?
The survey clarifies why so many educators welcome AI to help with assessment. But our second question remains:
Can AI deliver on educators’ expectations?
The answer here is a qualified yes.
Trent Workman, senior vice president of Pearson's School Assessment division, says AI can absolutely “make assessment more accessible and actionable” for educators.
Workman emphasizes the data piece. He’s not surprised that educators struggle to understand assessment data. One issue is that teachers must “pull data from different systems that don’t talk to one another. " Accessing the results takes more time. Worse, the measures have no standardization, so teachers have difficulty making sense of the data.
Workman concludes, “By providing an easy way to analyze various data points, AI can help educators tailor their instruction to better meet student needs and support school goals.”
The point of assessment is to tailor instruction. So, AI can indeed help educators—and students—by deriving actionable insights from assessment results.
The qualified yes
Why is the answer to the second question only a qualified yes?
One reason comes through in the survey. Most respondents voice concerns about the effectiveness of their current assessment tools. They need help to pinpoint student needs, measure progress toward learning targets, and personalize instruction. These aren’t data-analysis problems. The underlying issue is that the assessments don’t produce trustworthy results to inform instruction.
Another crucial reason is that many educators use AI to generate assessment questions rather than just score the answers. Uncritically embracing AI in this area can create more problems for teachers.
Assessments produce valid and reliable data only if they meet specific standards of rigor. The questions must follow principled assessment design practice, be aligned to state academic standards,grade-level appropriate, and free of biases.
AI-generated questions are not likely to meet these standards without some human review and intervention. Classroom teachers and school leaders alike must be aware of the limits of AI tools. They need high-quality assessment systems.
Otherwise, as Workman says, teachers rely on “AI-generated insights that are inaccurate or misleading.”
K-12 schools that don’t adopt AI to help with assessments could fall behind. But at the same time, those who go all in on AI may create insufficient data. And then they’ll hurt — not help — student learning.
To read more of the survey findings, download our report today.