Information Technology Reference
In-Depth Information
appear in two different flavours: with and with-out
a strong artificial intelligence component.
Theories incorporating a strong AI viewpoint
are inherently ill-defined as they need to take into
account all context variables that may influence
the learning process of the learners. Invested in
this approach, however, is a naive objectivist as-
sumption that it is possible to create an omniscient
artificial system that knows everything (or in a
weaker form 'everything important') about the
current context variables influencing a learner in
his information processing and learning work.
This is not possible. Learners are not sitting in
a glass-box where a teacher can monitor which
Wikipedia pages they are reading, who they are
talking to in the hallways, and whether their child-
hood experiences influence them towards reading
or watching television. Even if a learner could
have lived his whole life in a glass-box, still it
would not be possible to distinguish the relevant
environmental influences from the irrelevant ones
and the resulting representational model(s) would
have the same complexity as the original learner.
By itself it already would require an infinite
amount of adaptation work, which might even
grow exponentially with the number of people
participating in a learning network. And still -
like in Searle's famous Chinese room - even if
a system could number-crunch a problem of this
complexity, it would never truly understand what
the learner is thinking.
Contemporary instructional design theories,
however, have abandoned this goal of a strong
artificial intelligence monitoring and guiding
automatically a long time ago. Usually, they
foresee a mixture of minor automatic system
adaptations along a coarse-grain instructional
design master plan engineered by a teacher or
instructional designer. The so-called 'learning
paths' are fine-tuned along learner characteristics
and user profiles to conform to trails envisioned,
not necessarily proven by teachers.
There are two good reasons why these weak AI
theories have to be rejected for personalization.
First (and less important), there is no 'perfect'
instructional designer: an environment can only
be planned for the average learner, not the indi-
vidual. Even good instructional designers had to
gain their experience, had to make errors in the
past to built up effective and efficient strategies.
Moreover, in practice instructional designers are
most often 'only' domain experts for a particular
field of knowledge, no didactic experts. Second
(and more important), planned adaptation takes
experiences away from the learners: external plan-
ning keeps them from becoming competent, as it
takes chances to self-organize away and personal
discovery is prevented. Learners, however, are not
only sense-makers instructed by teachers along a
predefined path. Learners need to actively adapt
their learning environment to their needs so that
they can construct the competences necessary for
successful learning. And facilitators can coach
them along this way.
The emergence of a learning environment, in
the rich sense of interacting people-artifacts-tools,
is one (if not 'the') important outcome of a learn-
ing process, not just a stage to perform a 'learn-
ing play'. For these good reasons, we therefore
consider the better-known instructional design
theories to be flawed. Learners are not patients
that need an aptitude treatment. They proactively
have to (and of course already do) take account
of their learning environment.
Adaptation technologies can vary in their de-
gree of control: how much end-users are involved
in decisions about adaptation. Oppermann, Ra-
shev, and Kinshuk (1997) therefore differentiate
between adaptive and adaptable systems with a
fluent segue from the one to the other. Systems
are considered to be adaptable if the users initiate
the adaptation (and vice versa). Similarly, Dolog
identifies two perspectives through which adapta-
tion can be seen: adaptations can be performed
by humans to cope with changed requirements
of the participating stakeholders. Alternatively,
adaptation can be a dynamic system adaptation to
Search WWH ::
Custom Search