Season 2 of Sold a Story dropped today, and Emily Hanford opens with a claim about American education: a lot of people don’t believe instruction matters that much. Family, neighborhood, life circumstances, yes. Teaching, less so.

The season returns to 1966, when James Coleman’s congressionally ordered study found that differences in achievement were associated more closely with students’ families and neighborhoods than with the schools they attended. The Johnson administration released it quietly on a holiday weekend. Many read the findings as a limit on what schools could accomplish. Daniel Patrick Moynihan heard a challenge: perhaps a different kind of school could produce a different result.

Zig Engelmann was already trying. An advertising executive with no education background, he had been asked by a candy company how many repetitions it took for children to learn something from television. He couldn’t find the research, so he began conducting his own experiments. Eventually, he quit advertising. By 1966, he was teaching preschoolers from a poor Champaign, Illinois, neighborhood to solve basic algebra problems.

His colleagues describe reviewing the day’s lessons, identifying where children got confused, and rewriting the instruction. They added scripts to make the wording precise and give teachers more attention to spare for the children in front of them. APM’s accompanying explainer describes the substantial research base that later developed around Direct Instruction. But the working principle was there from the beginning: if the student hasn’t learned, the teacher hasn’t taught.

That care is hard to find in Lily Altavena’s reporting in Chalkbeat today. Teachers Pay Teachers plans to ban AI-generated classroom resources after Chalkbeat found apparent AI slop for sale, including an alphabet worksheet missing the letter F. Some sellers object, saying the crackdown threatens their income. Catching a missing letter is basic quality control. Establishing whether a resource teaches children anything takes more work.

That’s the work Sara Schapiro is asking Washington to fund. In Fast Company, the Alliance for Learning Innovation’s executive director asks where states, districts, and schools promising to “lead with evidence” on AI will get it. Education Week’s Lauraine Langreo reports that Stanford’s SCALE Initiative reviewed more than 800 papers on AI in K-12 and found only 20 with strong causal evidence. ALI’s recommendations include a standard the American Psychological Association has also asked for: assess students with the tool, then assess them again without it to find out what they learned.

Budgets are adding urgency. Victoria Feng reports in The Atlantic that AI’s demand for memory chips is driving up Chromebook prices, forcing districts to reconsider purchases and, in some cases, whether they can sustain a device for every student. Schools have to decide what deserves their money while researchers are still establishing what works.

Engelmann had a demanding definition of that word. Bonnie Grossen, who designed curriculum with him, recalled a lesson where 28 of 30 children learned the concept. She told him it worked.

He said it didn’t. There were still two children who hadn’t learned.

— Thomas

Looking for your next opportunity in education? W/A Jobs features 3,100+ career opportunities from 300+ organizations across the education industry.