What Ails U.S. Student Math and Reading Skills?

Some might claim educational technology is to blame for declining U.S. student capabilities in math and reading. That probably is not the most-common explanation, however. 

U.S. student performance in reading and mathematics has been tracked primarily through the National Assessment of Educational Progress (NAEP). Overall, scores showed steady improvements from the 1970s through the early 2010s, peaking around 2013. 

Since then, performance has stagnated or declined, with sharp drops during the COVID-19 pandemic (2019–2022) due to school disruptions. Test scores indicate high school scores for the 2024 graduating classes have dropped. 

Many reasons have been advanced, but some might point to a couple of decades of declining scores. 

Post-pandemic recovery has been uneven: math has shown slight rebounds in some grades, while reading continues to decline. 

These trends are more pronounced among lower-performing students, widening achievement gaps. By 2024–2025, 12th-grade scores reached historic lows, with reading 10 points below 1992 levels and math at its lowest since 2005.

Study/Author(s)	Year	Key Proposed Reasons
Scammacca et al	2019	Initial proficiency levels create Matthew effects (higher initial skills lead to faster growth); demographic factors (socioeconomic status, ethnicity) influence starting scores and growth rates; grade-level differences affect trajectories, with lower performers growing slower over time.
Wyckoff (Annenberg Working Paper)	2025	Pre-2013 gains slowed due to policy shifts (reduced accountability post-NCLB); Great Recession funding cuts; Common Core disruptions; rising smartphone/social media use; demographic shifts (more English learners); pandemic accelerated existing declines, especially for low performers.
Malkus (AEI Report)	2025	Similar to Wyckoff: accountability rollback, funding reductions, smartphone proliferation, and out-of-school factors (student well-being) explain stagnation since 2013; pandemic worsened trends but isn't the sole cause; factors outside school (e.g., screens) play a major role.
DeBord (Dissertation)	2026	Post-pandemic: classroom factors (rebuilding routines, teacher adaptability); school collaboration; family/district supports; persistent gaps in foundational knowledge, reduced struggle tolerance, technology habits, and socioemotional needs hinder recovery.
NWEA Brief	2021	COVID-specific: remote learning disruptions led to below-pre-pandemic gains; greater impacts on math than reading; inequities in access (devices, support) affected lower performers more; recovery requires addressing missed foundations.

But there may be lots of other cumulative reasons, including social promotion (advancing students to the next grade despite inadequate performance to maintain age-appropriate grouping) and efforts to avoid stigmatizing learners (through reduced retention or labeling to prevent emotional harm), may contribute to declining academic achievement in U.S. students. 

These practices are often adopted to mitigate short-term social-emotional risks like low self-esteem or dropout likelihood from grade retention, but research suggests they can exacerbate long-term performance issues by allowing students to advance without mastering foundational skills, leading to compounded learning gaps, disengagement, and lower scores on assessments like NAEP.

This is particularly evident among lower-performing students, where declines have been steepest since the early 2010s. It’s complicated.

AI Workloads Might Reach 50-70% of all Data Center Workloads by About 2030

By 2030, AI operations might represent half of all data center workload. But some analysts believe AI operations will account for as much as 70 percent of total data center workloads by 2030. 

source: McKinsey

At the very least, we might note that new data center construction (about 75 percent of new capacity)  now focuses on high-performance computing capabilities. 

High-performance computing sites might never number more than conventional data centers. But capability and “locations” are different metrics, as are “hyperscale” sized facilities and “typical” data centers. 

Layoffs: An Unfortunate Effort to Quantify AI Productivity Gains

Layoffs might be an unfortunate way of attempting to prove artificial intelligence productivity gains when there are few other ways to quantify the benefits in the near term. 

When important new technologies are introduced, there is almost always a lag between adoption and quantifiable productivity gains. In fact, it often happens that productivity drops as the new technology is adopted. 

Employees must take time away from current tasks to learn how to use the technology; test its results and so forth. 

There is no reason to expect the J curve of technology adoption will fail to be seen for, either. 

source 

Economic historians such as Erik Brynjolfsson and Paul David have documented that transformative, general-purpose technologies tend to follow the J-curve pattern. 

Initial deployment generates negative or flat productivity returns relative to investment, often for a surprisingly long time. 

David's famous 1990 paper on the "dynamo paradox" showed that electrification of US industry began in earnest in the 1880s but didn't produce measurable aggregate productivity gains until the 1920s.

The reasons are structural: firms must reorganize workflows, retrain workers, build complementary infrastructure, and abandon legacy processes before the technology's benefits materialize. 

The productivity gains, when they finally arrive, are real and large, but they accrue after enormous sunk costs and a long gestation period.

Amara's Law also suggests we will overestimate the immediate impact of artificial intelligence but also underestimate the long-term impact. But, again, that suggests it will be hard to quantify AI productivity results in the near term. 

All that is going to be a problem for financial analysts and observers who demand an immediate boost in observable firm earnings or revenue, as well as the firms deploying AI that will strive to demonstrate the benefit.

But layoffs are quite quantifiable, even if we might argue it is still too early to measure AI productivity impact.

Neither "AI Bubble Burst" Nor "Enterprise Software is Dead" Extremes Seem Likely

If you are an investor in either enterprise software; high-performance data centers or both, you face a truly-odd scenario, where, at the same time, we are warned that an “AI investment bubble” is underway at the same time that enterprise software (or software as a service generally) is at risk of extreme disruption from AI. 

One might argue software disruption or a high-performance computing “bubble” are conceivable, but not both together, as they tend to be mutually exclusive outcomes. 

In other words, the logical problem is that if AI really does not produce financial returns that justify the huge investments, then perhaps enterprise software actually is not badly disrupted.

On the other hand, if enterprise software really is disrupted, then the AI investments ought to have paid off, as they have proven consequential enough to destroy much of the value of enterprise software. 

And it always is possible that neither extreme scenario develops. In that case, perhaps high-performance computing facilities do produce a reasonable return, if perhaps not a “once in a lifetime” extraordinary return. 

And perhaps enterprise software adapts to AI, reshaping its business model successfully, though perhaps suffering a loss of profit margins in some instances. 

It’s the sort of scenario we often encounter. Some worry that Nvidia’s high margins or gross sales will be affected by competitors, ranging from AMD to in-house chip replacements from Alphabet or AWS, for example. 

And, often, there is some degree of displacement, but not ruin. And while we cannot know the timing or extent of disruption (so we must choose our “buy” prices), the extreme scenarios tend not to emerge. 

Netflix might not ultimately acquire Warner Brothers Discovery. Paramount might get the asset. Is it the “end” for Netflix? Not many would agree. Might Netflix have to find other ways to keep growing revenue? 

Yes. 

But apocalypse is not a likely outcome. 

Prudent investors will attempt to take advantage of market mispricing. And some degree of mispricing is likely. Volatility can be “your friend.” 

Scenario	The "Bear Case"	Why it invalidates the other Bear Case
AI Infrastructure Fails	Data centers sit empty; no ROI.	SaaS is Safe: If AI isn't useful enough to pay for compute, it isn't powerful enough to replace the complex workflows and "systems of record" that SaaS provides.
SaaS is Disrupted	AI agents replace apps and "per-seat" pricing.	Infra is Validated: To disrupt SaaS, AI must be doing the work. This requires massive, constant compute. The "Capex" wasn't a waste; it was the foundation of the new economy.
The "Middle Path"	AI becomes a feature, not a substitute.	Both Win/Lose: This is the boring reality where SaaS companies integrate AI and raise prices, and Infra gets a steady, if not "hyperscale," return.

We are never free from the risk of picking the “wrong” assets in a momentarily-depressed asset class, or making the wrong bets on whole categories of assets. 

But it seems to me unlikely the extreme “AI infra bubble” burst or “enterprise software is dead” theses will happen.

Some bad bets will be made, no doubt. That always happens. But do we really want to bet against the magnitude of impact?

Disintermediation, Again

Disintermediation, the removal of elements in a value chain particularly related to distribution, was a primary effect of the internet. 

Artificial intelligence should cause even more disintermediation, as processes and value chain roles dependent on information asymmetry are removed. 

source: WallStreetMojo

Worse, some might fear, there might be no natural restorative mechanism similar to the boom-bust; recession-recovery; supply-demand cycles we commonly see in economies. 

Instead, in a worst-case scenario, AI continually depresses consumer spending (which represents about 70 percent of all economic activity) as AI leads to layoffs, which leads to less consumer spending, which increases the necessity of relying on AI to protect firm profit margins. 

Industry	How AI Removes Friction/Asymmetry	Expected Impact
Real Estate Brokerage	AI agents instantly access MLS data, decades of transaction records, valuations, and matching—eliminating agents' knowledge advantage and buyer/seller search friction.	Commissions drop sharply (e.g., from 5-6% to under 1%); many deals close agent-free or via AI "agent-on-agent"; widespread disintermediation and value loss for traditional firms.
Insurance Brokerage/Underwriting	AI enables instant policy comparison, re-shopping, risk prediction via vast data, and fraud detection—eroding inertia-based renewals and broker expertise.	15-20% premium loss from passive renewals; brokerage spreads compress; shift to direct/AI-driven models; broker selloffs (e.g., triggered by tools like Insurify).
Wealth Management / Financial Advisory	AI delivers personalized portfolio advice, tax strategies, and real-time market analysis—democratizing what once required expensive human experts and proprietary insights.	Erosion of 1% AUM or high advisory fees for basic services; "basic financial advisory" faces collapse; robo-advisors and AI tools dominate routine work.
Legal Services (Routine)	AI automates contract drafting, precedent research, due diligence, and basic navigation of laws—reducing asymmetry between lawyers and clients on standard matters.	Commoditization of routine work; reduced demand for junior/entry-level roles; faster/cheaper services pressure billable hours and firm margins.
Travel Booking Platforms	AI agents autonomously assemble custom itineraries (flights, hotels, etc.) faster/cheaper than platforms, bypassing search friction and default inertia.	Margin compression for intermediaries; habitual booking models disrupted; platforms lose value as direct AI routing prevails.
Logistics / Freight Brokerage	AI optimizes routes, matches shippers/carriers directly, and forecasts with superior data—eliminating broker coordination friction and information edges.	Broker fees collapse; rapid selloffs (e.g., truck brokers like RXO); shift to automated direct platforms.
Consulting / Professional Advisory Services	AI handles research, data analysis, and initial recommendations—eroding human expertise moats and proprietary knowledge asymmetry.	Reduced fees for routine/cognitive tasks; white-collar headcount cuts; productivity gains but revenue destruction for traditional models.
Ratings Agencies / Index Providers / Credit Checks	AI aggregates and analyzes public/proprietary data at scale—undermining exclusive research advantages and manual verification friction.	Proprietary edges erode; commoditization of ratings/indexing; lower barriers and pricing pressure.

Payments also are an obvious place to look for changes, though the changes might come from use of blockchain-based payment processors offering stablecoins.

source: Citrini Research