Hesham Khalafallah
When Learning Began to Leave a Trail
In a traditional classroom, the teacher notices many things.
They see the student who raises a hand, the one who looks lost, the one who answers quickly, and the one who needs another explanation.
But some things remain hard to see, especially in a large class.
Two students can get exactly the same grade even though their journeys were completely different.
One may have understood from the first explanation.
The other watched the lesson three times.
One got the main concept wrong before correcting the mistake.
The other may have memorized the answer without really understanding it.
The final result may be similar.
But the learning that came before it is not.
Digital environments began to make part of that journey visible.
From the Grade… to the Learning Journey
For a long time, we summed up a student’s performance in a single number.
70%.
85%.
Excellent.
Very good.
Fail.
But a number does not always tell us everything.
One student may score 80% because they have mastered almost every concept except one essential skill.
Another may get the same score with an average level across the whole subject.
The numbers are identical.
The teaching decisions they call for are not.
This is one reason education is paying more attention to ideas such as skills mastery — Mastery — rather than the final grade alone.
Instead of saying:
The student scored 75%.
We can try to say:
They have mastered the first skill.
They still need help with the second.
They are progressing quickly in practice.
And they keep making the same mistake on one specific concept.
A grade tells us how many answers the learner got right, but it does not always tell us what they are actually able to do.
What Does “Learning Analytics” Mean?
The term Learning Analytics may sound very technical, but the basic idea is simple.
We use the data produced during learning to better understand what is happening to the learner, and then we try to use that understanding to improve the learning experience.
This data can include:
Time spent on an activity.
The number of attempts.
The questions the student got wrong.
The lessons they went back to.
The activities they completed.
Their progress across different skills.
And sometimes even how they interact inside a digital environment.
The goal should not be to collect data simply because it exists.
The goal should be to turn data into better teaching decisions.
Data does not become educational when we collect it; it becomes useful when it helps us make a decision that improves learning.
Can the Teacher Know Sooner That a Student Needs Help?
This may be one of the most important possibilities learning analytics offers.
In the traditional model, we sometimes discover a difficulty only after the exam.
But what if the teacher could see it while it is happening?
A dashboard might show that a group of students stopped at the same point.
Or that many of them made the same mistake.
Or that one student has not finished several activities in a row.
Another may be moving very quickly and need an extra challenge.
In this case, data does not replace the teacher.
It can become a tool that widens their ability to observe.
Data should not replace the teacher’s eye; it should help them see what is hard to notice among dozens of students.
The relationship between the teacher and the class changes once again.
They no longer have to wait only for the final result.
They can step in during the journey.
From the Same Education for Everyone… to Different Support for Each Learner
If we know that a student needs to review one specific concept, why ask them to repeat the whole lesson?
And if we know that another has already mastered the idea, why hold them back?
This is where data begins to support what we discussed in the article on artificial intelligence:
personalization.
One student may get an extra explanation.
Another a different example.
A third more practice.
A fourth a more advanced challenge.
The goal stays the same.
The path can change.
The better we understand the learner, the more possible it becomes to give them what they need instead of giving everyone exactly the same thing.
Differences between learners are no longer a problem to be eliminated.
They become information that can help us design better learning.
When Artificial Intelligence Meets Learning Data
This is where this article connects directly with article 17.
There is a big difference between an AI that knows only the question I am asking it now and an AI built into an education system that knows part of my earlier journey.
If the system knows that I keep running into the same difficulty, it can suggest a review before I move on.
If it knows that I am progressing quickly, it can offer a harder activity.
If it notices that I memorize answers but cannot apply them, it can change the type of activity.
Personalization becomes deeper.
AI becomes better at personalizing when it knows not only the learner’s current question, but also understands something about their earlier journey.
But this important possibility immediately raises another question:
How much do we want the system to know about the learner?
From the Platform to the Immersive Environment… Data Goes Deeper
In the previous article, we talked about extended reality and simulation.
There is a big difference between knowing that a student finished a lesson on a platform and knowing what they actually did in a virtual lab.
In an immersive environment, the experience can leave much more detailed traces.
Which route did they choose?
Which device did they head to?
Which decision did they make first?
Where did they go wrong?
How many times did they repeat the experiment?
When did they ask for help?
Did they manage to complete the task on their own?
Data then comes closer to behaviour inside the experience, not just the time of logging in or out.
This opens new possibilities for assessing practical learning.
We no longer only know whether the learner got the right result.
We can try to understand how they got there.
Can Assessment Itself Change?
This is a very important question.
In the traditional model, much of the assessment comes at the end.
We teach.
Then we test.
But if the environment can observe the learning journey continuously, assessment can also become more continuous.
The system does not necessarily need to wait for a big exam to understand that a student is struggling.
The difficulty may be visible through ten small attempts made during learning.
We can then gradually move from the question:
What grade did the student get in the exam?
to a broader question:
What can they do now that they could not do before?
The best evidence of learning is not always the moment we ask the student to show what they know; sometimes it is being able to see how their skill develops during the process itself.
But Is Everything We Can Measure Important?
This is where the problem begins.
Digital environments can measure many things.
The number of clicks.
Time.
Speed.
The number of attempts.
The completion rate.
Interaction.
Attendance.
But do all these numbers mean that we truly understand the learner?
Not necessarily.
One student may spend a long time because they are thinking deeply.
Another may spend exactly the same time because they are distracted.
One may rewatch a video because they did not understand.
Another because they enjoyed the explanation.
One student may stop an activity because of an Internet problem.
Another because they lost motivation.
Data can often tell us what happened.
It does not always tell us why it happened.
Data can teach us a great deal about the learner’s behaviour, but it cannot always explain why they behaved that way.
This is where the human comes back in.
The teacher understands the context.
The family knows the child’s circumstances.
And the learner knows what they are going through.
Can a Student Become a Digital Profile?
This question worries me.
The more capable systems become at building learner profiles, the easier it becomes to label learners too early.
This student is weak in mathematics.
This one is slow.
This one is not engaged.
This one is brilliant.
But human beings change.
Children change even more.
Data describing a student’s performance today must not become a final judgment on what they will be capable of tomorrow.
Data should help us open possibilities for the learner, not build a digital box in which we lock them.
There is a big difference between saying:
This student needs help now.
and saying:
This student is weak.
The first sentence describes a need.
The second can become an identity.
How we talk about data therefore becomes part of educational ethics in itself.
Is the Algorithm Fair?
If the system uses data to recommend a level, a path or an activity, another question arises:
Is the decision fair?
What if the data is incomplete?
What if the student has a poor connection?
What if the platform’s language is not their first language?
What if the activity itself does not suit their needs?
The algorithm may observe poor performance.
But the real reason may lie outside the learning process.
So we must not let an automated decision become a final verdict that cannot be questioned.
The more capable the system becomes of making decisions, the more important it is for people to understand why those decisions were made and to be able to review them.
This concerns the teacher.
The learner.
And the family.
Who Owns the Learner’s Data?
This may be one of the most important questions for the future of education.
If the platform knows:
The student’s level.
Their mistakes.
Their difficulties.
Their learning history.
Their behaviour in the environment.
And sometimes more…
who owns this data?
The school?
The platform?
The technology company?
The student?
The family?
Who can access it?
How long is it kept?
Can it be deleted?
Can it be used for purposes other than learning?
These are not only technical questions.
They are also educational and ethical questions.
The ethical question is not only: what data can we collect? It is also: what data do we actually need to collect to improve learning?
Not everything that can be measured should be measured.
And not everything that can be stored should be kept.
Is a Child’s Data the Same as an Employee’s?
For me, the caution must be even greater when children are involved.
A young learner may not fully understand what it means to consent to having their data collected.
They may not know where it will be sent.
Or how it might be used later.
That is why the design of educational technology must not start only with the question:
What can we know?
It must also ask:
What do we actually have the right to know?
This places a great responsibility on schools, developers, policymakers and families.
Can Data Become a Lifelong Memory of Learning?
In the previous article, we talked about Web3, digital identity and Verifiable Credentials.
A new idea appears here.
What if a learning record were not limited to final certificates?
What if a person had a record showing the skills they mastered at different stages of their life?
Learning at school.
Then at university.
Then on a platform.
Then in a company.
Then in a professional programme.
Then in a simulation.
Then through self-learning.
Such a record could become much richer than a single degree earned at twenty-two.
This brings us back to what we said in article 14:
Being a learner is no longer just one stage of life.
Learning can become a lifelong journey, accompanied by an evolving picture of the skills and experiences gained.
But this opens another question:
Who controls this memory?
Data Should Not Run Education on Its Own
Numbers have something seductive about them.
They look precise.
They look neutral.
They create a sense of certainty.
But education is more complex than a dashboard.
The system may say a student did not finish the activity.
The teacher may know they are going through a difficult family situation.
The system may say the student is slow.
The teacher may discover that they think deeply.
The system may say a student is performing well.
But the teacher may know that they cannot work with others and need to develop a completely different skill.
Data can widen our understanding of the learner, but it cannot reduce a whole human being to a screen.
This is the balance we must protect.
Perhaps the Teacher Will Rely Less on Impressions Alone
At the same time, we should not fear data simply because it is data.
Teachers can make mistakes too.
They can form an inaccurate impression.
They may not notice a quiet student in a large class.
They may believe everyone understood simply because no one asked a question.
Data can become an additional mirror.
It is not the whole truth.
But it can offer another angle.
The teacher needs data, and data needs the teacher: the first reveals patterns, the second gives them meaning and context.
The future relationship then becomes clearer.
It is not:
The human or the algorithm.
But:
The human with the algorithm.
What Happens When We Understand Each Learner Better?
If we can use data responsibly, we can come closer to a very old dream of education.
A teacher who knows when a student needs help.
A challenge suited to each student.
A difficulty discovered before it turns into failure.
Progress seen in ways other than the exam.
Feedback given at the right moment.
More flexible learning.
But these benefits will not come from collecting the largest possible amount of data.
They will come from collecting the right data, interpreting it well, and using it through responsible human judgment.
Smart education does not mean knowing everything about the learner; it means knowing enough to help them without losing respect for their privacy and their humanity.
From “Education for All”… to “Understanding Every Learner”
In the early days of mass education, one of the great achievements was simply giving large numbers of people access to schooling.
Then the question became:
How do we improve quality?
Then:
How do we make education available everywhere?
Then:
How do we personalize it?
Today a new stage is emerging:
How do we better understand the learner themselves?
Not to watch them more closely.
But to support them more precisely.
There is a big difference between a system that says:
“Here is the same lesson for everyone.”
and a system that says:
“The goal is shared, but here is the help you need right now.”
But a Human Being Is Bigger Than Their Data
After everything we have discussed, I come back to a principle I consider essential.
A system may know thousands of things about a student.
But it may not know their dream.
It may know that they stopped at the seventh question.
But not know why they are afraid of failure.
It may know that they did not finish an activity.
But not know that an encouraging word from their teacher made them come back the next day.
It may know how fast an answer was.
But it does not necessarily know what success means to this human being.
This reminds us of something that may become even more important as technology advances:
The more the system knows about the learner, the greater our responsibility not to forget that behind the data is a human being who cannot be reduced to data.
We May Now Be Nearing the End of the Journey
We began this series with one human being learning from another.
Through observation.
Through storytelling.
Then came writing.
The book.
The school.
The curriculum.
Sound and image.
The computer.
The Internet.
Distance learning.
Global platforms.
Social networks.
The smartphone.
Artificial intelligence.
Extended reality.
And now the system is beginning to see part of the learner’s journey through data.
But after all these tools, a bigger question remains.
If the machine knows more…
explains more…
measures more…
and predicts more…
what remains for the human being?
Perhaps here we reach the last question of this series:
Humans Learn (20): From Storytelling to Artificial Intelligence… What Will Remain Human?
Because at the end of the journey, the question is no longer only:
How will learning tools continue to evolve?
It becomes:
What kind of human being do we want all these tools to help shape?
