EdTech Leader, University Adjunct, Consultant, Podcaster, & 11x Author matthewrhoads.com flow.page/drmattrhoads
San Diego, CA
- Tweets 11,821
- Following 6,211
- Followers 5,818
- Likes 32,152
Pinned Tweet
Super excited to have received the notification from @SolutionTree that #CoTeachingEvolved has been published. This upcoming week pre-orders will be fulfilled and on October 11th, the book will release on Amazon.
Learn more: coteachingevolved.com
A thread...
#SolutionTree
2
2
10
838
The strongest AI finding so far is not that AI helps students answer. It is that unrestricted access hurts transfer while timed access does not. When students use AI matters more than whether they use AI.
2
389
Practice testing beats rereading. Spacing beats cramming. Neither costs a dollar. Before any district buys another platform, the question is whether it makes those two things happen more often. If it does not, it is not instructional.
447
IBM surveyed 2,000+ K-12 educators and parents: 76% of middle school educators use AI weekly, only 20% have extensive training. We do not have an adoption problem. We have a preparation problem. What would training need to cover for you?
2
596
Most districts wrote an AI policy before writing a curriculum. The policy governs the tool. The curriculum governs the learning. One of those was the wrong priority.
439
Knowledge is what the mind thinks with. A student who knows nothing about the topic cannot think critically about it, no matter how good the prompt is. Critical thinking is not a skill you drill. It is a dividend of knowing things.
1
440
One teach, one observe is the most common co-teaching model in American schools. It is also the one where the second teacher disappears. If the observation never feeds back into instruction, you are paying two salaries for one lesson.
192
A new Stanford review analyzed 800+ studies on AI and K-12 education. The number of high-quality causal studies was close to zero. The next vendor who shows you a efficacy deck: ask for the study design. If there is no control group, there is no claim.
149
The Kindle edition of Crush It from the Start is out today. I wrote it for the coach responsible for other adults' growth with no authority over them.
Get it: amazon.com/Crush-Start-Instr… #ad
1
292
PISA 2025 has a result almost nobody in edtech is sharing: moderate classroom technology use is associated with better student performance than heavy use. The dose makes the poison. Where is your line?
928
PISA 2025 dropped and the conversation went straight to AI. The quieter finding: US reading scores fell 14 points. You cannot comprehension-strategy your way out of a knowledge deficit. What are students actually reading this year?
132
The testing effect is the most replicated finding in learning science, and the least used. Students who practice recalling outperform students who reread, every time. AI tools make generating practice questions trivial. So why is PD still about the tool?
114
Most co-teaching will move toward AI-mediated differentiation, not parallel teaching. We see the shift now: the AI handles the tier, the teachers handle the relationship. The model is the access point. What will AI-mediated differentiation look like in your classroom?
680
AI-assisted students answered 48 percent more problems correctly. They also scored lower on the transfer test. The tool helped them do the problem. It did not help them own the skill. I keep this study in mind because the gain is real and the transfer is the question. When did.
1
2
406
Cognitive load is the enemy of the new year. Every new thing a student must hold at once is a load. I tell teachers to subtract before they add. The lesson that works is the one with the fewest things to hold. What will you remove from your first unit to lower the load?
1
611
A 30-point AI gap inside the same district: 76% of middle school educators use AI weekly, only 45% of elementary. K-5 needs a seat in the AI guidance conversation.
newsroom.ibm.com/2026-09-02-…
1
373
Districts are buying AI tools faster than researchers can study them. That is the real risk, not the tool. I tell leaders to buy the evidence with the tool, not just the tool. When did your district last check the research behind an AI purchase?
3
499
The summer buying rush is dead: 54% of districts buy on a rolling basis. But 57% still have no significant AI procurement policy.
Where does your district stand on AI procurement?
smartbrief.com/original/the-…
68
76% of middle school educators use AI weekly. Only 20% have extensive training on it. Adoption isn't the bottleneck, readiness is.
What's the AI training situation at your school?
newsroom.ibm.com/2026-09-02-…
1
701
76% of middle school educators use AI weekly. Only 20% have extensive training on it. Adoption isn't the bottleneck, readiness is.
What's the AI training situation at your school?
Source: newsroom.ibm.com/2026-09-02-…...
87