4 responses

  1. Mark Gbur (@thestaticfrost)
    February 5, 2013

    The Adjacent Possible of an Adaptive Learning Platform from the University of Phoenix

    I am very intrigued by the University of Phoenix Adaptive Learning Platform that Phil Hill wrote about, but have reservations that “data analytics is going to kill the standardized curriculum dominant in higher education.” That assumption seems predicated on an opinion that an environment could replace, rather than merely enhance self directed learning. Ultimately, I think self directed learning has a limit, an adjacent possible.

    Out-of-the-gate, adaptive learning poses hurdles to a learner framed in one or two manners, it can evolve with time and continuous improvement but often times initial arrangement is too elementary (think of trying to setup a system to automatically graded a fill-in-the-blank question; there are numerous ways one can construct a response of “22222.23 mL” in a string, inclusion for scientific notation, variations of expressing units, and so forth). Similarly, users can hit a roadblock when posed with two iterations of a question and need guidance that an adaptive system could not provide.

    I find that adaptive technology is not always relevant and presently, is too slow, it is too far behind “what is now.” Facebook, which has “learned about me” and solicits posts it thinks I want to read often is bringing up popular content amongst my social graph, however popular does not translate into relevant. My true interests are much more narrow than my social graph and I could care less about the majority of social gossip, click bait, and hollow grandstanding that encompasses the most interaction on the Facebook. I actually have to tailor the experience on the system quite a bit just to drown out the noise. I had an easier time chronologically navigating through content in the initial configuration because I had an inherent logical control for that noise. More importantly, curated content is too slow, just look at Flipboard. Things surface by what is popular, and what is popular takes accumulation and time. If I want new, breaking information in technology, my best bet is to go to a human edited curation site like Techmeme, not wait for Flipboard to tell me what is automatically curated by popularity.

    My worry about dependence on an Adaptive Learning Platform is that often times, true understanding comes from radically different proposals of others rooted in experiences outside of a framed context. I believe that we will still need to rely heavily on SMEs who have established experience to frame rationale and theory in culturally relevant and socially iterative frameworks. Beyond simply establishing an Adaptive Learning Platform, it is equally important to foster environments where learners can grow on their own as it is to be able to provide personalized insight rooted in experience, both with regard to the problem and the context.

    The adjacent possible is human edited curation. We still need our SMEs to tell us what is important, frame it in the experience of the field, root it in the context of life, and simply recognize a disconnect that would constitute a series of variables at-present, only the human brain is capable of accumulating, analyzing, and adjusting for to reach desired outcomes. The Adaptive Learning Platform is next, but it will take many iterations, time, and close care by SME’s monitoring outcomes to achieve independence. Data analytics will be transformitive, but we learn in context, from those we care about and relate to, and in experience, there is still a very human element.

    Cheers,
    -Mark Gbur
    @thestaticfrost

  2. Michael Staton
    February 6, 2013

    This project will bomb. All good developers know you need five or less people and a safe space to iterate. Expensive projects by large bureaucracies almost always fail.

  3. Phil Hill
    February 6, 2013

    Mark, your points are well taken, and if you have not done so, I strongly recommend reading Michael’s post: http://mfeldstein.com/why-big-data-mostly-cant-help-improve-teaching/

    The one issue take is your meta-assumption that “That assumption [big data to replace standardized curriculum] seems predicated on an opinion that an environment could replace, rather than merely enhance self directed learning.” Look at the work starting at San Jose State, where they’re taking an edX MOOC, and applying it to a flipped classroom. Also look at some of the implementations of Knewton. These efforts are looking to enhance self-directed learning, or at least small-group-directed learning. I don’t believe every big-data / adaptive learning approach is looking to replace rather than enhance.

    And my descriptions are not to say that these efforts (specifically Phoenix’s) will succeed. I’m describing their attempts and the massive scope involved.

    Thanks for comment.

  4. Phil Hill
    February 6, 2013

    Michael, their effort certainly does go against the grain of agile development and small-team innovation theories. It will be interesting to watch.

    When I get time, I plan to analyze more on the results we’re starting to see, which might shed light on whether the project will bomb or not. I would note ahead of time, however, that their promised deployment of the platform has already changed dates and the renewed investment might indicate some challenges.

Back to top
mobile desktop