Note: These notes are based on a presentation
at the NSF Symposium “Learning & Intelligent Systems,” June 26th, 1996.
They are structured in accordance with the major questions and issues discussed
at the symposium. The author would like to thank

(1) the organizers of this symposium for the opportunity
to present the ideas and the research efforts behind them documented in
these notes, and

(2) his collaborators at CU Boulder, specifically the
members of the L³D Center who contributed in the development
of the ideas behind these notes.

Feedback to these notes would be greatly appreciated
(gerhard@cs.colorado.edu).

More information about the L³D Center can
be found at: http://l3d.cs.colorado.edu/

Table of Contents

• 1. The State of Relevant Knowledge
• 1.1 Learning: Current Theories
• 1.2 Lifelong Learning
• 1.3 Lifelong Learning and Design
• 1.4 Beyond the “Gift Wrapping” Approach
  of Educational Reform – Rethinking, Reinventing, and Reengineering Education
• 2. Important Frontiers for Fundamental
  Interdisciplinary Research
• 2.1 Myths and Misconceptions
• 2.2 Hypotheses and Requirements for
  Computational Environments Supporting Lifelong Learning
• 2.3 The Grounding of Learning and Intelligent
  Systems in Theories of Learning
• 3. Building a Successful Interdisciplinary
  Investigation
• 4. Responses of the Relevant Science,
  Engineering Research, and Education Communities
• 4.1 Challenges
• 4.2 What’s Wrong With Current Universities
  
• 5. Long-Term Societal Impacts of This
  Research

Next Generation Authoring Tools & Instructional
Applications

Introduction. There is general agreement as we
approach the next century and next millennium that our society is changing
into a knowledge and information society. We will face new opportunities
and new challenges in all dimensions of our lives. But the future is not
out there to be “discovered”: It has to be invented and designed. A research
agenda on “Learning and Intelligent Systems” must focus on “making learning
a part of life,” and the implications this has on how‹under the influence
of new media, new social structures, and new objectives for a quality
of life‹human beings will think, create, work, learn, and collaborate
in the future.

1. The State of Relevant Knowledge

1.1 Learning: Current Theories

Current trends in educational theory make the following
fundamental assumptions about learning (arguments supporting this view
can be found in the books by L. Resnick “Knowing, Learning and Instruction”,
D. Norman “Things That Make Us Smart”, and M. Csikszentmihalyi “Flow”):

• Learning is a process of knowledge construction, not of knowledge
  recording or absorption.
• Learning is knowledge-dependent; people use their
  existing knowledge to construct new knowledge.
• Learning is highly tuned to the situation in which
  it takes place.
• Learning needs to account for distributed cognition
  requiring knowledge in the head to combined with knowledge in the world.
  
• Learning is affected as much by motivational issues
  as by cognitive issues.

1.2 Lifelong Learning

Lifelong Learning: A Ubiquitous Goal. Lifelong learning
has emerged as one of the major challenges for the worldwide knowledge society
of the future. A variety of recent events support this claim: (1) 1996 is
the “European Year of Lifelong Learning,” (2) UNESCO has included “Lifetime
Education” as one of the key issues in its planning, and (3) the G7 group
of countries has named “Lifelong Learning” as a main strategy in the fight
against unemployment. Despite this great interest, there are few encompassing
efforts to tackle the problem in a coherent way. Lifelong learning cannot
be investigated in isolation by looking just at one small part of it, such
as K-12 education, university education or worker re-education.

Learning as a New Form of Labor. The previous
notions of a divided lifetime-education followed by work-are no longer
tenable. Learning can no longer be dichotomized, spatially and temporally,
into a place and time to acquire knowledge (school) and a place and time
to apply knowledge (the workplace). Professional activity has become so
knowledge-intensive and fluid in content that learning has become an integral
and inseparable part of “adult” work activities. Professional work can
no longer simply proceed from a fixed educational background; rather,
education must be smoothly incorporated as part of work activities fostering
growth and exploration. Similarly, children require educational tools
and environments whose primary aim is to help cultivate the desire to
learn and create, and not to simply communicate subject matter divorced
from meaningful and personalized activity.

Lifelong learning is a continuous engagement in acquiring
and applying knowledge and skills in the context of authentic, self-directed
problems. L³D’s theoretical framework for lifelong learning
is grounded in descriptive and prescriptive goals such as: (1) learning
should take place in the context of authentic, complex problems (because
learners will refuse to quietly listen to someone else¸s answers to someone
else¸s questions); (2) learning should be embedded in the pursuit of intrinsically
rewarding activities; (3) learning-on-demand needs to be supported because change is inevitable,
complete coverage is impossible, and obsolescence is unavoidable; (4)
organizational and collaborative learning
must be supported because the individual human mind is limited; and (5)
skills and processes that support learning as a lifetime habit must be
developed.

Figure 1: Education & Technology

1.3 Lifelong Learning and Design

Lifelong learning integrates and mutually enriches the cultures of work
and education. Central to this vision in our own research is the notion
of design activity, a model of work that is open-ended and long-term in
nature, incorporates personalized and collaborative aspects, and combines
technical and aesthetic elements. Design is an argumentative
process, involving ongoing negotiations and trade-offs; it is also a collaborative
process making increasing use of new social structures brought about by
the advent of computer networks and “virtual communities.” The communality
that crucially binds these and other design activities together is that
they are centered around the production of a new, publicly accessible artifact.
Engineers and architects design infrastructure and buildings, lawyers design
briefs and cases, politicians design policies and programs, educators design
curricula and courses, and software engineers design computer programs.
It is impossible for design processes to account for every aspect that might
affect the designed artifact. Therefore design must be treated as an evolutionary
process in which designers continue to learn new things as the process unfolds.
The relationship between learning and design provides the impetus for the
work done at the L³D Center. Because design
is an essential aspect of all problem-solving activity, and since designers
are constantly learning and communicating with each other, the research
done at the L³D Center seeks to ground
educational theory within the domain of technology that supports design
and communication.

1.4 Beyond the “Gift Wrapping” Approach
of Educational Reform-Rethinking, Reinventing, and Reengineering Education

Figure 2: The “Gift Wrapping” Approach

A deeper understanding and more effective support for
lifelong learning will contribute to the transformation that must occur
in the way our society works and learns. A major finding in current business
reengineering efforts is that the use of information technology had disappointing
results compared to the investments made in it (see for example the book
by Tom Landauer “The Trouble with Computers”). While a detailed causal
analysis for this shortcoming is difficult to obtain, it is generally
agreed that a major reason is that information technologies have been
used to mechanize old ways of doing business‹rather than fundamentally
rethinking the underlying work processes and promoting new ways to create
artifacts and knowledge.

Figure 3: Rethinking & Reinventing Education

We claim that a similar argument can be made for current
uses of technology in education: it is used as an add-on to existing practices
rather than a catalyst for fundamentally rethinking what education should
be about in the next century. For example, the “innovation” of making
transparencies available on the World-Wide Web (WWW) rather than distributing
copies of them in a class takes advantage of the WWW as an electronic
information medium. This may change the economics of teaching and learning,
but it contributes little to introducing new epistemologies. “Old” frameworks,
such as instructionism, fixed curriculum, memorization, decontextualized
learning, etc., are not changed by technology itself. This is true whether
we use computer-based training, intelligent tutoring systems, multimedia
presentations, or the WWW. As Figure 1 shows, education often follows
on the boot heels of technology rather than guiding the appropriate development
and use of technology.

We need computational environments to support “new” frameworks
for education such as lifelong learning, integration of working and learning,
learning on demand, authentic problems, self-directed learning, information
contextualized to the task at hand, (intrinsic) motivation, collaborative
learning, and organizational learning. Figure 2 illustrates the “gift-wrapping”
approach in which technology is merely wrapped around old frameworks for
education. Figure 3 indicates what is needed instead: a richer conceptual
framework, leading not just to the addition of technology but to the weaving
of technology into learning and working.

Figure 4 tabulates the major changes required. It shows
strong similarities between the behaviorist learning theory of B.F. Skinner
and the models of industrial work of F.W. Taylor, and contrasts these
with the lifelong approach to learning.

2. Important Frontiers for Fundamental Interdisciplinary
Research

2.1 Myths and Misconceptions

The current debate about the ability of computation and communication to
fundamentally change education are (in our opinion) based on a number of
fundamental myths and misconceptions. The most prevalent ones are:

• Computers by themselves will change education-
  There is no empirical evidence for this assumption based on the last
  30 years of using computers to change education (such as computer-assisted
  instruction, computer-based training, or intelligent tutoring systems).
  Technology is no “Deus ex machina” taking care of education. As mentioned
  before, making slides available over the World-Wide Web rather than
  giving paper copies to students can be valuable, but will not change
  education. Instructionist approaches are not changed by the fact that
  information is disseminated by an intelligent tutoring system.

• Information is a scarce resource -“Dumping”
  even more decontextualized information on people is not a step forward
  in a world where most of us already suffer from too much information.
  Instead, technology should provide ways to “say the ‘right’ thing at
  the ‘right’ time in the ‘right’ way.” In our research, we have explored
  problems associated with high-functionality applications (such as operating
  systems, word processors, spreadsheets, etc.). Our empirical findings
  (which are universally true for all systems) are illustrated in Figure
  5. These systems provide challenging problems for a research agenda
  for “Learning and Intelligent Systems,” because if future “progress”
  is achieved only by extending D₄ to D₄‘, there
  will be no benefits for users. Instead of increasing the tool mastery
  burden of users even more, we need new concepts such as learning-on-demand,
  information delivery, and task-based unfolding,
  so users can incrementally explore and master such systems according
  to their needs.

The rectangle (D₄) represents users’ knowledge about the system’s
information
space. D₁ represents concepts well known and easily employed
by the users.
D₂ contains concepts known vaguely and used only occasionally,
often requiring
passive help systems. D₃ represents concepts users believe
to exist in the system,
some of which lie outside the actual information space. In the case of
increased
functionality (as illustrated by D₄), the area D₄-D₃
(representing the functionality
users are not even aware of) increases to D₄‘ – D₃,
not that of the ovals.
Figure 5: Levels of Users’ Knowledge About a
System’s Information Spaces

• The content, value, and quality of information
  and knowledge is improved just because it is offered in multi-media
  or over the WWW-Media itself does not turn irrelevant or erroneous
  information into more relevant information (as indicated by Figure 6).
  We must create innovative technologies (such as simulations, visualizations,
  critiquing, etc.) to let people “experience” knowledge in new ways.
  

Figure 6: The Existence of Information Alone is Not Good Enough

• “Ease of use” is the greatest challenge or the
  most desirable goal for new technologies -Usable technologies that are not useful for the needs and concerns
  of people are of no value. Rather than assuming people should and will
  be able to do everything without a substantial learning effort, we should
  design computational environments that provide a low threshold for getting
  started and a high ceiling to allow skilled users to do the things they
  want to do.

• The “Nobel Prize Winner” myth: Every school child
  will have access to a Nobel Prize winner-This was one of the selling
  points for the information superhighway. While this argument is true
  (or will be true soon) at the level of technical connectivity, it is
  doubtful that Nobel Prize winners will look forward to getting a few
  thousand e-mail messages a day.

• The single or most important objective of computational
  media is reducing the cost of education -Although we should not
  ignore any opportunity to use technology to lessen the cost of education,
  we should not lose sight of an objective that is of equal if not greater
  importance: increasing the quality of education.

• Human learning is equal to machine learning -Although
  we have deepened our understanding of human learning through progress
  in machine learning, there are fundamental dimensions, such as motivation
  and competing requirements for a human¸s time, that make human learning
  a much more complex and interwoven activity than machine learning. There
  is substantial empirical evidence that the chief impediments to learning
  are not cognitive. It is not that students cannot learn; it is that
  they are not well motivated to learn.

2.2 Hypotheses and Requirements for Computational
Environments Supporting Lifelong Learning

H1: User-directed and supportive. In any computational
system supporting lifelong learning, the choice of tasks and goals (including
the learning opportunities offered) must be under the control of the user/learner,
and support must be contextualized to the user¸s task.

H2: Contextualized presentation. A system supporting
lifelong learning must present information to the user in a way that is
maximally relevant to the user’s chosen project or task.

H3: Breakdowns as opportunities for learning. A system supporting
lifelong learning will be sufficiently open-ended and complex that users
will encounter breakdowns. The system must provide means for allowing
users to understand, extricate themselves, and learn from breakdowns.
Rather than attempting to eliminate trouble, the system should help users
manage troubles and exploit breakdowns as opportunities rather than failures.

H4: End-user modification and programmability. A
system supporting lifelong learning must provide means for significant
modification, extension, and evolution by users.

H5: Supporting a range of expertise. Systems supporting
lifelong learning will be employed over long periods of time by their
users; hence, these systems must be able to accommodate users at progressively
different levels of expertise.

H6: Useful and usable. Many existing research
efforts and computational environments reflect an implicit belief and
are grounded in a design philosophy that there is an inevitable design
tradeoff between the notions of usefulness and usability. Systems supporting
lifelong learning must be useful and usable.

H7: Promoting collaboration. Systems supporting
lifelong learning must include means for collaboration among users.

2.3 The Grounding of Learning and Intelligent
Systems in Theories of Learning

Technologies Transcending the “Gift-Wrapping” Approach.
The grounding of technologies in new theories of learning (as briefly summarized
in section 1.1) will contribute to transcend the “gift-wrapping” approach
as shown by the following attributes of learning environments:

• Learning is a process of knowledge construction
  requiring environments in which learners can be active designers
  and contributors rather than passive consumers. Research in end-user
  programming and end-user modifiability contributes toward this goal.

• Learning is knowledge-dependent requiring environments
  supporting user-tailored information presentations such as differential
  descriptions of new information (for example: if someone wants to learn
  HTML and knows MS-WORD, the explanations and examples provided should
  be different than those given to a learner who knows Framemaker).

• Learning is highly tuned to the situation in which
  it takes place requiring environments which are domain-oriented
  and which support human problem-domain interaction and not just human computer interaction. The information
  spaces presented and the information provided should be made relevant
  to the task at hand, something which computational media can achieve,
  but which is impossible for paper and pencil technologies.

• Learning needs to account for distributed cognition
  requiring environments which create and define new role distributions
  between humans and computers. Most of what any individual “knows” today
  is not in her or his head, but is out in the world (e.g., in other human
  heads or embedded in media).

• Learning is affected as much by motivational issues
  as by cognitive issues requiring environments which let people experience
  and understand why they should learn and contribute something. For example,
  learning-on-demand lets users access new knowledge in the context of
  actual problem situations and delivers information about which they
  are unaware in the context of their problem situations. Environments
  must allow users to take pride in their contributions and be awarded
  for them.

From Knowledge Aquisition to Knowledge Construction
and Evolution: The Seeding, Evolutionary Growth and Reseeding (SER) Model.
Most intelligent systems (including systems in support of learning such
as Intelligent Tutoring Systems and Expert Systems) of the past have been
developed as “closed” systems. The basic assumption was that during design
time, a domain could be modeled completely by bringing domain experts
(designers) and environment developers (knowledge engineers) together
and the knowledge engineers would acquire the relevant knowledge from
the domain experts and encode it into the system. This approach fails
for the following reasons: (1) as argued before, much knowledge is tacit
and only surfaces in specific problem situations; and (2) the world changes,
and intelligent systems modeling this world must change accordingly. In
our research, we have developed a process model to address these problems
(see Figure 7). It postulates three major phases:

A seed will be created through a participatory design process
between environment developers and domain designers. It will evolve in
response to its use in new design projects because requirements fluctuate,
change is ubiquitous, and design knowledge is tacit. Postulating the objective
of a seed (rather then a complete domain model or a complete knowledge
base) sets this approach apart from other approaches in intelligent systems
development and emphasizes evolution as the central design concept.

Evolutionary
growth takes place as learners use the seeded environment to undertake
specific projects. During these design efforts, new requirements may surface,
new components may come into existence, and additional design knowledge
not contained in the seed may be articulated. During the evolutionary
growth phase, the environment developers are not present, making end-user
modification a necessity rather than a luxury. World-wide communities
of practice can participate in this process if the WWW becomes an information
environment for collaboration and sharing rather than one for information
dissemination.

Reseeding, a deliberate effort of revision and coordination
of information and functionality, brings the environment developers back
to collaborate with domain designers to organize, formalize, and generalize
knowledge added during the evolutionary growth phases. Organizational
concerns play a crucial role in this phase. For example, decisions have
to be made as to which of the extensions created in the context of specific
design projects should be incorporated in future versions of the generic
design environment. Drastic and large-scale evolutionary changes occur
during the reseeding phase.

Figure 7: The SER Model: A process model for the development and
evolution of domain-oriented intelligent systems

3. Building a Successful Interdisciplinary
Investigation

Building successful interdisciplinary investigations is not a small task
in a world in which specialization necessarily increases and the days of
the universally educated “Renaissance Scholars” belong to the past. C.P.
Snow, in his famous book “The Two Cultures,” identified the difficulty of
“literary intellectuals” and “natural scientists” communicating successfully
with each other. He claimed to have found a profound mutual suspicion and
incomprehension, which had damaging consequences for the prospects of applying
technology to the alleviation of the world¸s problems. Many more different
cultures exist today, e.g., novices versus skilled workers, software developers
versus software users, industry people versus academics, and committed technophiles
versus determined technophobes.

Experiences. At CU Boulder we have tried for the
last ten years to build bridges among different cultures (the most relevant
ones will be briefly mentioned):

1. The Institute of Cognitive Science at CU Boulder brings
together researchers from the humanities, the social sciences, the natural
sciences, and engineering, acknowledging that problems the scientific
community needs to address do not always fall neatly into the structures
of established departments.

2. In the context of university/industry relationships,
we have tried to reinvent the purposes of such collaborations (and have
explored research issues in detail in our close collaboration with NYNEX
University and NYNEX Science and Technology).

3. By working with the Boulder Valley School District
and with several specific schools, we have tried to understand the problems
of empowering teachers to become lifelong learners and of introducing
and sustaining technology in school settings.

4. Acknowledging that learning is desired and takes place
outside formal institutions, we have recently started a collaboration
with the Boulder County Healthy Community Initiative, a group of several
hundred concerned citizens, which reflects on the future of our county.

5. In our L3D Center we have brought together researchers
and students from various parts of the world to understand different perspectives
how people think about our world. Our focus on lifelong learning and design
has served as a forcing function to create these interdisciplinary investigations
and they in return have been of critical importance to our understanding
of the challenges of learning and intelligent systems.

Challenges. The building of successful interdisciplinary
investigations faces the following challenges:

1. To regard the existing “symmetry of ignorance” (a concept articulated
by Rittel, who argues that among all the carriers of knowledge for any
real problem there is nobody who has a guarantee that her or his knowledge
is superior to any other person’s knowledge) as an opportunity rather
than as a limitation or an undesired obstacle.

2. To overcome the boundaries of creating divisions between
basic and applied research by doing basic research on real problems.

3. To find ways and to develop means to allow different
cultures to talk to each other and to engage them as active participants
in inventing the future (e.g., to liberate social scientists from their
passive consumer and Cassandra role, and to make technologists aware that
technological changes and innovations do not happen in isolation but in
existing social networks involving people).

4. To make sure that in multidisciplinary approaches
to learning and intelligent systems all relevant disciplines are contributing
(e.g., one might argue that at this time the design of learning and intelligent
systems can profit more from research of social psychologists on matters
of motivation, creativity, role models, etc. than from neuropsychology).

4. Responses of the Relevant Science, Engineering
Research, and Education Communities

4.1 Challenges

“Making Learning a Part of Life” creates many challenges, requiring creative
new approaches and collaboration among many different stakeholders. For
illustration, just a few of them will be mentioned here.

1. The educated and informed citizen of the future:
‘super-couch potato’ consumers or enlightened designers-The major innovation
that many powerful interest groups push for with the information superhighway
is to have a future where everyone shows her or his creativity and engagement
by selecting one of at least 500 TV channels with a remote control. The
major technical challenge derived from this perspective becomes the design
of a “user-friendly’ remote control. Rather than serving as the “reproductive
organ of a consumer society” (Illich), educational institutions must fight
this trend by cultivating “designers,” i.e., by creating mindsets and
habits that help people become empowered and willing to actively contribute
to the design of their lives and communities. This goal creates specific
challenges for computational artifacts, such as the support of end-user
programming and authoring.

2. The “basic skills” debate-If the hypothesis
that most job-relevant knowledge must be learned on demand is true, we
have to ask ourselves: What is the role of “basic skills”? If, for example,
the use of software packages dominates the use of mathematics in the workplace,
shouldn¸t a new function of mathematics education be teaching students
to use these mathematical artifacts intelligently? Another important challenge
is that the ?ld basic skillsº such as reading, writing, and arithmetic,
once acquired, were relevant for the duration of a human life; modern
?asic skillsº (tied to rapidly changing technologies) will change over
time.

3. Can we change motivation?-As mentioned, there
is substantial empirical evidence that the chief impediments to learning
are not cognitive but motivational. This raises the challenge of whether
we can create learning environments in which learners work hard, not because
they have to, but because they want to. We need to alter the perception
that serious learning has to be unpleasant rather than personally meaningful,
empowering, engaging, and even fun. In our research efforts we have developed
computational environments to address these motivational issues; for example,
our systems have explored making information relevant to the task at hand,
providing challenges matched to current skills, creating communities (among
peers, over the net), and providing access to real practitioners and experts.

4. School-to-work transition-If the world of working
and living (a) relies on collaboration, creativity, definition, and framing of problems; (b) deals with uncertainty,
change, and distributed cognition; (c) copes with symmetry of ignorance;
and (d) augments and empowers humans with powerful technological tools,
then the world of schools and universities needs to prepare students to
function in this world. Industrial-age models of education and work (based
on Skinner and Taylor, as illustrated above) are inadequate to prepare
students to compete in the knowledge-based workplace. A major objective
of our lifelong learning approach is to reduce the gap between school
and workplace learning. Our research addresses some of the major “school-to-work”
transition problems and develops answers to the following questions:

• How can schools prepare learners and workers for a
  world that relies on interdependent, distributed, non-hierarchical information
  flow and rapidly shifting authority based on complementary knowledge?

• What “basic skills” are required in a world in which
  occupational knowledge and skills become obsolete in years rather than
  decades?

• How can schools (which currently rely on closed-book
  exams, the solving of given problems, and so forth) be changed so that
  learners are prepared to function in environments requiring collaboration,
  creativity, problem framing, and distributed cognition?

• To what extent will lifelong learning and new approaches
  to learning and teaching-such as learning on demand, learning while
  working, relations, and the involvement of professionals in schools‹prepare
  learners for work?

4.2 What’s Wrong With Current Universities

We consider the self-application of our theories a critical element (and
a unique opportunity) in the assessment of our research efforts. Universities
as institutions need to be in the middle of rethinking the future of working
and learning-applying their findings not only to other institutions, but
to themselves. Using the previously developed framework causes us to critically
examine our own work as university faculty members in the following ways:

• Understanding learning as active knowledge construction
  rather than passive knowledge absorption questions the dominance of
  lectures.

• Allowing learners to engage in authentic, self-directed
  learning activities is at odds with micro-managed curricula.

• Acknowledging that problem solving in the real world
  includes problem framing calls into question the practice of asking
  students to solve mostly given problems.

• Recognizing that most interesting problems in the real
  world do not have right or wrong answers, but instead must be solved
  by satisfying objectives that are most important for that situation.

• Acknowledging that the individual human mind is limited
  and that outside of schools people rely heavily on information and knowledge
  distributed among groups of people and various artifacts (distributed
  cognition) questions the value of closed-book exams, and requires a
  much greater emphasis on collaborative learning and communication skills.

5. Long-Term Societal Impacts of This Research

Research in learning and intelligent systems, especially if we want to move
beyond the “gift wrapping” approach of technology, will have fundamental
long-term societal impacts. It will force us to reinvent how we think, work,
learn, create, and collaborate. It will change

1. institutions, e.g.,
   
   • universities (as argued above)
   • companies will have to become learning organizations

2. individuals, e.g.,
   
   • who will have a desire to become independent of
     high-tech scribes in personally meaningful and important activities
     
   • . who would like to contribute to their (computer-enriched)
     reality rather than merely interacting with it

3. mindsets, e.g.
   
   • teachers should see themselves not as truth-tellers
     and oracles, but as coaches, facilitators, learners, and mentors
     engaging with learners
   • breakdowns and symmetry of ignorance need to be
     understood as opportunities
4. connections and collaborations, e.g.,
   
   • connecting in new ways (e.g., distributed communities
     of practice and interest) will go along with disconnecting in old
     ways (being physically together, increased specialization)
   • organizational learning supported by organizational
     memories will complement individual learning.

This research will provide us with opportunities to explore fundamentally
new possibilities and limitations of computational media as they complement
existing media. It will force us to think about new concepts such as sustainable
communities of practice. It will pose the question of how large complex
information spaces can be evolved over long periods of time, not by their
professional designers but by their affected users. It will enrich the notion
of distributed cognition, allowing us to draw different lines between what
humans should do and what machines should do.

One may argue that our current thinking does not address
the potential magnitude of the change. Have we arrived at a point where
the change is of a similar magnitude to the time when our society moved
from an oral to a literary society (and Socrates and Plato were arguing
about the trade-offs associated with this change) or when Gutenberg’s
printing press eliminated the scribes and gave everyone the opportunity
to become literate? The fact that societies have often overestimated change
in the short run and underestimated it in the long run suggests that we
should make every effort to understand the long-term societal impacts
of learning and intelligent systems.

Conclusion. As argued at the beginning, the future
of how we live, think, create, work, learn, and collaborate is not out
there to be “discovered” it has to be invented and designed. Computational
and communication media (firmly grounded in a deep understanding of theories
and prescriptive goals) will be a critical force in shaping this future.
Learning and intelligent systems research will play an important role
in creating this future.