Amanda Bickerstaff's Recent LinkedIn Posts

A new report on 9-17 year olds' use of AI by Common Sense Media puts a spotlight on the widening gap between AI usage and AI literacy. While 85% of students who use AI turn to it for schoolwork, a large majority lack even the most basic knowledge of how AI works. The report found that out of the respondents who said they know "a lot" about how AI works, only 37% knew that GenAI can't tell the difference between what's true and what's false. Perhaps most concerning is the finding that students who struggle academically, including those who have a hard time staying focused, lean on AI the most: 56% compared with 46% of their peers. These students are turning to AI for help, but without AI literacy, their use runs the risk of exposing them to misinformation, overreliance, and cognitive offloading. Schools have a critical role to play. Policies and guidelines are a start, but there is a danger in only communicating to students what not to do. Effective AI literacy teaches students how to make informed decisions not only about how to use GenAI, but also about when they should set the tools aside. Our SEE GenAI Literacy Framework lays out the importance of AI literacy instruction that counters misconceptions, builds foundational knowledge of how the tools work, and cultivates critical mindsets. This is how students move from confident but uncritical AI use to true self-efficacy, where their use is grounded in safe, ethical, and effective practices. Link to the full report in the comments. AI for Education #GenAI #AILiteracy

I was asked by the 74million to review a new Stanford study that followed ~350 elementary students across two districts who were given access to an AI reading tutor. Here is what they found. Left to use it on their own, most students barely touched it with nearly half never logging on at all. Across all of the students, weekly use averaged just 2-5 minutes, far below the 30 minutes the developers of the tool say kids need before they see any benefit. Pairing students with a human tutor for check-ins and encouragement had a very small impact. Usage went up by 1-4 minutes a week. In both cases, the AI tutor didn't improve their reading scores, which tells us very little, since the tool barely got used in the first place. None of this surprised us as we've seen the pattern before. Sal Khan, one of the earliest champions of AI tutoring, has admitted that Khanmigo was "a non-event" for many students who simply didn't use it. While it's always good to have new research into the impact of AI EdTech tools, I'm not sure this one is even asking the right question. The question we should be grappling with shouldn't be how to get young kids to use an AI reading tutor more often. Measuring usage was the easy part. The better question is whether these tools belong in early elementary classrooms at all. Study and articles linked in the comments. AI for Education #GenAI #aisafety #K12 #teachingwithAI

Most of the conversation about AI risk in learning has focused on what happens to our thinking when we offload it. New research suggests we should be just as concerned about what happens to our relationships. Researchers at the University of Oxford and Stanford ran five studies with over 3,000 participants, including a three-week study where people worked through personal dilemmas with a chatbot. The research focused on "sycophantic" AI, or the tendency for AI to affirm and validate users rather than push back. Some highlights from the research: 🔹After talking with sycophantic AI, participants perceived human conversations as requiring more effort. 🔹Over three weeks, participants became nearly as likely to seek advice from AI as from close human connections, and they found human social interactions less satisfying. 🔹Importantly: AI interactions felt good in the moment but didn't deliver the tangible benefits afterward that come from human connection. This mirrors what we see with learning and AI: it can often feel productive but not lead to a lasting positive impact. The researchers describe sycophantic AI as offering people the experience of being understood "but without the work that produces it." This means listening, vulnerability, and empathy: the work that builds and sustains real relationships. This research has me thinking about findings from Common Sense Media's "Talk, Trust, and Trade-Offs" survey from 2025. That survey reported that 31% of teens found conversations with AI companions as satisfying as, or more satisfying than, those with their friends. Neither piece of research shows people actively pulling away from other humans. Instead, the shift was perceptual: participants didn't spend less time with the people in their lives, they just found that time less satisfying. The most sobering finding from the research is that choice didn't fix it. When participants tried all three AI styles and picked one to keep talking to, a majority chose the sycophantic version, not because its advice was better, but because it felt easiest and most understanding. The researchers concluded that giving users style options won't be enough, and that mitigation has to happen at the model level. That leaves young people, families, and schools navigating tools designed to be the easiest conversation in the room unless OpenAI, Google, and Anthropic make significant changes to the way they build and train models. Link to the original study in the comments. #ailiteracy

What does it actually take to bring high quality GenAI literacy training and adoption strategies back to your school or district? Join us in AI for Education's next webinar tomorrow at 12pm ET for an open conversation about what this work looks like in practice. We will hear from alumni of our three micro-credential courses as they share the policy frameworks, staff PD programs, and AI literacy training resources they built during our programs, and discuss what happened when they implemented their projects in their organizations. This sessions is for anyone that wants to hear directly from practitioners leading AI adoption efforts in their own schools and districts. In this session, we will hear: ✅ What it looks like to develop a district AI policy and implementation strategy that fits your community ✅ How school leaders have structured AI adoption, including sequencing teacher and student literacy, and addressing grading and academic integrity ✅ How instructional coaches designed staff PD that connects AI literacy directly to instructional planning ✅ What works best for AI literacy training targeted at parents and caregivers Link below to learn more or register. #AILiteracy #K12 #education

In our workshops, we keep seeing the same thing: when students recieve good AI literacy training means more critical use of the tools. GenAI is designed to feel effortless. You ask a question and get a fluent, confident answer in seconds. That ease is the trap. Good GenAI use takes real oversight, and most students are operating in the dark. A new study out of France (Clerc et al., 2026) shows what changes when you teach that oversight. The researchers gave 116 middle schoolers a set of science tasks to work through with a GenAI tool. Beforehand, 76 students attended a two-hour AI literacy workshop; the other 40 did not. The workshop wasn't just some prompting pointers. It covered how LLMs work and why they fail (hallucinations, bias, sycophancy) along with a few transferable practices like adding more context to prompts and asking follow-up questions. The results showed that trained students: 🔹 Were more likely to reject weak, underspecified prompts. 🔹 Asked follow-up questions to get better responses more often (59% vs. 28%). 🔹 More accurately judged whether GenAI responses were correct. One other takeaway the evidence keeps surfacing: how confident students felt about AI didn't predict how well they used it. While this is a small study with modest gains, it is an initial proof point that short GenAI literacy training can teach students better, more critical use of the tools. You can check out the study in the comments. AI for Education #AIliteracy #K12