I have been fascinated by Artificial Intelligence (AI) for years. Like nuclear energy, which can be used for good or evil, AI is both promising and scary: a genie we MIGHT want to put back in a bottle.
Today’s NYTimes “Bits” section, which offers insights into the latest developments in technology, describes how a Canadian company is using AI developed in the US to “personalize” shoe purchases by providing customers with an ever narrowing range of options that fulfill their immediate desires. Here’s a description of how it works:
Customers browse pictures of shoes, choosing a favorite type among a dozen images, which leads to a dozen more images, searching for the look someone is after. An initial page of boots might lead to a page with more low-cut items over high, or laces over buckles. The next options could affirm those choices. Or the search could go off in another direction.
From a sales point of view, it could be considered the next step in the A.I. of personalization. Typically, personalization relies on historic associations, or the familiar “customers also bought” suggestion. In this case, the computer is looking at 100 or more factors, and trying to judge how someone feels about them in real time.
“What makes this attractive is that people can get to the shoe they love without knowing what brand it is,” said Roger Hardy, chairman and co-founder of Shoes.com. “If I told you there was an Italian company with the perfect heel, toe and lacing for you, but didn’t know the brand, it wouldn’t do you any good.”
In early customer tests, Mr. Hardy said, the A.I. increased sales, though he declined to say by how much.
An earlier paragraph in the article led me to think of ways this kind of AI application might be used in education:
While (this AI application) is — at least for the moment — limited to retail, over the long haul the technology could demonstrate how important it is for companies to be sitting on vast warehouses of information.
Of late several articles have been written about the vast trove of electronic records we are keeping on students and the fact that some school districts have unwittingly allowed this data to be used for commercial purposes. However, as I’ve noted in earlier posts on this issue, public schools have been keeping written records on children for decades, records that could arguably provide teachers with data on a student’s learning patterns in the same way. Using the highlighted sentence above and making a few substitutions, here’s the result:
…personalization in the classroom relies on a student’s optimal learning modality by using an analysis of 100 or more factors that reveal previous classroom successes when a student was struggling to master a concept and linking them to the how student feels about this learning obstacle in real time. As a result, the teacher is able to see what approach they might take to intervene.
Is the development of such an algorithm for learning possible? If it were, it would clearly benefit schools who, like the shoe companies cited above, sit on “..vast warehouses of information”. And… if the presumed goal of personalization is to customize learning experiences to help each child define their goals and provide teachers with the tools to meet the needs of each child, why wouldn’t we use the “..vast warehouses of information” we’ve been collecting for decades to the best effect possible?
Harvard Business Review Synthesizes Peter Drucker’s Perspectives on Automation: Life-Long Learning is Essential
Roughly 20 years ago I read several books on management theory and found that almost all the writers on this topic based much of their thinking on the ideas of Peter Drucker, a prolific and insightful writer and thinker. Rick Wartzman, the Executive Director of the Drucker Institute in Claremont, CA wrote an essay for the Harvard Business Review titled “What Peter Drucker Had to Say About Automation” that synthesized his perspective on that topic, and they ultimately boil down to one idea: life-long learning is essential for any worker who wants to avoid becoming obsolete.
In the essay, Wartzman cites Drucker’s writings from a 1946 Harpers essay on the mechanization of cotton harvesting and concluding with quotes from his 1993 book The Post-Capitalist Society. In 1946 Drucker wrote:
“It is easy—and very popular in the Deep South today—to see only one aspect of the technological revolution through which the Cotton Belt is passing: the removal of the dead hand of the cotton economy and plantation society, the establishment of a sound agriculture and of a better balance between industry and farming, higher incomes, better living standards, the end of sharecropping—in short the final emancipation of both white and colored from slavery. It is also easy to see only the other aspect: dislocation, the suffering, the uprooting of millions of people who will lose their homes and their livelihood.
However, the full picture, as in all technological revolutions, emerges only if both—the better life for those who can adjust themselves and the suffering of those who are pushed out—are seen together and at the same time.
In 1986, nearly four decades after observing the impact of the mechanization of cotton harvesting, Drucker observed the same phenomenon in the rust belt:
The “shrinkage of jobs in the smokestack industries and their conversion to being capital-intensive rather than labor-intensive, that is, to automation, will put severe strains—economic, social, political—on the system,” Drucker warned in his 1986 book The Frontiers of Management.
In 1993, nearly five decades later, Drucker underscored the need for everyone to adapt to technological advances by learning new and different skills, envisioning a role for both traditional schooling and corporations:
“School,” Drucker wrote in 1993’sPost-Capitalist Society, “has traditionally been where you learn; job has been where you work. The line will become increasingly blurred.”
Employers also have their role, including “active and energetic attempts at retraining for specific new job opportunities,” as Drucker put it. And each employee must step up and be ready to embrace what’s being taught—over and over and over again. “People have to learn how to learn,” Drucker advised. “No one is allowed to consider himself or herself ‘finished’ at any time.”
The highlighted sentences from Post-Capitalist Society resonate with me the most, and should be the basis for determining if a high school graduate is “ready to work”. If “learning how to learn” is the ultimate goal of schooling, the use of a single test to determine if a student is “ready to work” or “ready for higher education” is preposterous. Passing a test implies that the student is a “finished” product. The ability to “learn how to learn” cannot be measured with a single test. It requires initiative, independence, and insight, traits that a teacher can observe and, if given the chance, could document. The ability to pass a test measuring a prescribed set of skills requires compliance and conformity. It doesn’t require a teacher’s observation, only the grading by a machine. Compliance and conformity might land a job, but being able to remain in the workplace in the future requires a combination of initiative and the ability to learn independently. And only a skilled teacher can motivate a student to want to learn… but instead of training and valuing teachers who can motivate independent learning we are trying to replace teachers with computers that train compliant students to pass tests.
Another year, another study and no solutions are in sight because only those with the least political muscle and the most need are the ones who suffer the most. As reported in Education Week as well as other news outlets, the third annual CoSN Infrastructure Survey reports that gaps persist in schools and in homes with– no surprise– rural and less affluent school districts and homes having less access to broadband and wifi access than urban, suburban, and affluent school districts and homes. And why is this the case? The second bullet in the Education Week summary by Benjamin Herold has the answer:
- Cost: Money is still seen as the biggest barrier to robust school connectivity: 46 percent of survey respondents identified the cost of monthly recurring charges as their biggest problem, followed by 34 percent of who cited high upfront capital costs.
E-rate funds, which were expected to help close the digital divide, are being challenged because fewer and fewer homes are using land lines, a factor that the survey indicated would affect 90% of the districts who responded to the survey.
This persistent technology disparity contributes to the persistent disparity between students raised in poverty and those raised in affluent homes as children and schools without readily available wi-fi cannot engage in innovative practices like the flipped classroom or research assignments that rely on internet searches as opposed to leafing through outdated encyclopedias. So the affluent suburban schools issue I-pads to each child while economically challenged schools send home worksheets… and we wonder why the performance gap persists.
Diane Ravitch had several posts yesterday on the deficiencies of Outcome Based Education, posts that yielded several strong dissents based on B.F. Skinner’s theories, computer-based individualized instruction, and early attempts at outcome based and self-paced education that relied heavily on handouts. I remain convinced that until we abandon our current mental model of education as one based on lockstep progression based on age based cohorts we will remain stuck in the same arguments I’ve witnessed for the four decades I’ve worked in public school administration.
We’ve used OBE based on common standards for decades in one area that requires students to demonstrate mastery with both academic and performance assessments… and a brief history of the delivery of this content in this discipline might shed some light on this issue and also on the direction public education could be headed.
Everyone in our country who possesses a drivers license passed both an academic assessment (typically a multiple choice test) and a performance assessment (typically an over-the-road review with a police officer). The standards a student must master in order to obtain a driver’s license are universal. The time required to master the academic and performance skills varies widely. Students who fail the assessments can re-take them as many times as is needed, but once an individual masters the skills as measured by the written and performance tests they receive a license that is no different from anyone else’s. Students who received the training in a structured program offered by a certified instructor received an additional benefit: insurers rewarded the completion of such a program with a reduced rate because their data showed that such students experienced a lower accident rate.
Students used to receive training for these OBE assessments in public schools but in most states the responsibility for learning these skills has shifted out of school and into the private sector. The rationale for this shift was two-fold: the cost for providing the equipment needed for training was high and the insurance benefits that resulted from the attainment of certificate would enable parents to fund the program out-of-pocket instead of having the program funded by taxpayers.
When public schools dropped Drivers Education, private drivers education schools proliferated. Some of the schools were staffed by former certified teachers whose compensation ad benefits were lower than those offered by public schools and others were staffed by instructors with credentials determined by insurance companies. Oh… and some of the students who might have experienced the financial benefits of taking a publicly funded course lost the opportunity to do so because their parents could not afford the out-of-pocket costs associated with enrolling in a privately operated school operated by an accredited teacher. Most of them DID get their drivers license but paid an insurance premium for several years thereafter.
I trust that readers of this blog can see how this brief history of drivers education might apply to the trends in public education we are witnessing today… and might highlight the consequence of our obsession with having everyone learn at the same pace. Because we accept the current model of schooling we fail to ask some basic questions:
=>Why do we group students in grade levels based on their age?
=>Why do we group students within a particular grade level based on their rate of learning?
=>Why do we group students at all?
=>Why does school take place in a limited time frame?
=>Why do we believe there is “one best way” to educate ALL children?
All of these practices are in place because they result in “efficiency” in the factory school… and until we change our minds about how schools are organized, until we replace our conception of schools as a factory with a new mental model, we will continue measuring “quality” by giving standardized tests to students grouped in “grade levels” and recycling “new ideas” and “reforms” based on ways to run the factory more efficiently.
Several years ago public schools decided to outsource the attainment of the drivers license “badge”. The “badges” being developed by private sector enterprises (e.g. IT certifications) are superseding the “diplomas” on the back end of the factory. How long before the same phenomenon occurs in public schools?
Over the past several weeks I’ve read several articles in technology related journals and progressive education blogs on the obsolescence of the diploma and its replacement with “badges” that signify “mastery” as opposed to “coverage”. As one who has long advocated mastery learning and was drawn to the idea of having a series of badges replace traditional diplomas, I offered the drivers license as a prime example of how badges might work. One obtains a divers license “badge” by demonstrating a combination of abstract knowledge (the passing of a pencil and paper test) with applied knowledge (the passing of the performance test). This same method of signifying “mastery” could (and I would argue should) be applied to every form of learning.
With that context, I found “The Rise of Micro-Schools”, a post on the Steelemaley blog, intriguing and promising. The Micro-schools described in the post would probably be categorized as “un-schooling” or “de-schooling” since there is no evidence that adults are creating learning modules. Instead, the students in the international micro-schools share a passion for ecology and instead of only reading about it, they are gathering data that scientists can use to inform their research, witnessing ecology first hand, working collaboratively in face-to-face situations and virtual international teams, and putting their findings and ideas into practice. While the post doesn’t say so explicitly, it is evident that the students are working with the guidance and support of adults as opposed to the direction and monitoring of adults.
Could this work for all children? It could if we determined that the purpose of schooling was to create young adults who are capable of working independently with minimal supervision. As long as we sustain our factory model of education, the one designed to churn out compliant employees who will work within defined frameworks, we will undervalue independence and interdependence and persist in our beliefs that anything else is impossible. Without saying so explicitly, blogger Steelemaley is indicating it’s time to challenge the dominant paradigm.
The NYTimes Mokoto Rich’s report on the decline in mathematics scores on the National Assessment of Educational Progress is full of explanations for the drop, but all of the explanations miss the primary factor. Here are the expert’s explanations, in the sequence they appear in the article, and my commentary on each in bold red
…it could be related to changes ushered in by the Common Core standards, which have been adopted by more than 40 states. For example, some of the fourth-grade math questions on data analysis, statistics and geometry are not part of that grade’s guidelines under the Common Core and so might not have been covered in class. The largest score drops on the fourth-grade math exams were on questions related to those topics.
This is highly implausible since many states had not rolled out these standards
…stagnating performance could also reflect the demographic changes sweeping America’s schools and the persistent achievement gap between white students and minorities, as well as between students from poor families and their more affluent peers
This is undoubtedly an underlying factor, particularly the divide between affluent and poverty-stricken schools. But this is not the primary factor.
…with students taking so many other standardized tests, some educators said those who took the national exams, which were administered from January to March, may simply have had test fatigue.
This is implausible. The NAEP is only administered to a small group of students within a district and when it was administered in districts I led it was very low key.
Protests about testing as well as decisions by some parents to opt their students out of testing could have influenced some students who took the national exams.
This completely bogus idea was presented by the Council of Chief School Officers, who are clearly grasping for straws. In order for this to be the case large numbers of parents of high achieving students would need to pull their child from school on the days when NAEP was administered… and unlike the high stakes tests given by States the NAEP has a wide window and is not highlighted in the minds of parents and children.
Of all the explanations, the AFT President came closest to nailing the real problem:
Randi Weingarten, the president of the American Federation of Teachers, linked the drop in test scores to recent educational policies as well as the economic downturn and its aftermath. “Of course we are disappointed” with the scores, she said. “But they should give pause to anyone who still wishes to double down on austerity and make competition, scapegoating teachers, closing rather than fixing schools, driving fear, and testing and sanctioning the dominant education strategies.”
The test-and-punish model implicit in NCLB and exacerbated by RTTT combined with austerity measures imposed on public schools is a close second to the real factor… which is how the focus of teachers’ energy changed when testing became the predominant metric for school success and teacher success.
A quick primer on test construction is needed to see why the teacher’s focus changed. NAEP is a norm-referenced examination, scored by determining the average (mean) test scores. The “high stakes tests” introduced by NCLB are criterion referenced tests whose results are based on the percent of students who score higher than a proficient level that is determined by a (arguably subjective) cut score. NCLB moved toward criterion referenced tests because on a norm referenced test the scores of high achieving students can offset the lower scores attained by struggling students, causing the lower scoring students to be “left behind”.
When NCLB’s criterion referenced testing was introduced it was clear what the consequence would be: teachers would focus most of their time and energy on students whose scores were just below the proficient level the prior year. While those children had been “left behind” when mean scores were the basis for determining if a school was succeeding… the focus on that group of students resulted in another group being “left behind” or, more accurately, “held back”: those in the top end of the achievement level. And when the higher achieving students were under-served, NAEPs mean scores were suppressed.
The solution? I remain convinced that if children were allowed to advance in reading and mathematics at their own pace that norm-referenced scores based on age cohorts would rise and, over time, more students would achieve the benchmarks set for high school graduation. As stated repeatedly in this blog, we have the technology available that can make this happen, we have the human resources in our schools that can make this happen, all we need to do is abandon our idea that students progress uniformly through their schooling based on their age, an idea we hold onto despite decades of evidence that it is absolutely wrong.