If AI Does the Beginner’s Work, How Will Our Children Ever Become Experts?
HappyKay Learning | AI, Education & Future Careers
If AI Does the Beginner’s Work, How Will Our Children Ever Become Experts?
At first, what caught my attention was simple:
AI is reviewing legal contracts.
Crosby describes itself not simply as a software company selling AI to lawyers, but as an AI-native law firm. Its AI agents process contract work while experienced lawyers remain involved for legal judgment and accuracy. The firm says it stands behind its work and carries malpractice insurance.
Crosby says that when it launched in 2025, its median contract review time was 58 minutes. By 2026, it was experimenting with things such as simulated responses to contract redlines and voice agents for negotiations.
Source:
Crosby — Introducing the Hybrid AI Law Firm
Crosby — Series B and Its AI-Native Law Firm Model
At first I thought:
Wow. Legal work is really changing.
Then another thought came to me as a mother.
Wait. If AI does the beginner’s work, how does a beginner ever become an expert?
That question bothered me much more than whether AI would “replace lawyers.”
What did young lawyers used to do?
Traditionally, a new lawyer might spend enormous amounts of time doing work that did not look particularly exciting.
Reading documents.
Searching case law.
Comparing contracts.
Reviewing discovery materials.
Writing research memos.
Preparing first drafts.
Then a senior lawyer would review the work.
The junior lawyer might discover that half of the draft had been rewritten.
That probably did not feel wonderful.
But something important was happening during those years.
The young lawyer was slowly learning to notice things.
Why did my senior lawyer delete this argument?
Why did she ask about this fact when I thought it was unimportant?
Why did he immediately become concerned when he saw this sentence in the contract?
Why did my research lead me in the wrong direction?
Thousands of little corrections gradually became something we call professional judgment.
The American Bar Association has raised this exact concern about AI and lawyer training.
Some of the inefficient and repetitive work traditionally performed by young lawyers provided what it calls “productive struggle”—the practice through which lawyers developed reasoning, judgment and professional instincts.
If AI removes too much of that experience, law firms and law schools will have to intentionally replace it with something else.
Source:
American Bar Association — The Future of Lawyer Training in the Age of AI
Here is the strange part.
We are beginning to tell young professionals:
“AI can do the basic work. Your value will be judgment.”
But I found myself asking:
How can a beginner have judgment without first having experience?
A 25-year-old attorney cannot simply wake up with the instincts of a lawyer who has handled hundreds of cases.
Yet AI may increasingly produce the first research, first analysis and first draft.
Then the junior lawyer may be asked:
Is the AI correct?
That sounds efficient.
But it creates a strange reversal.
The least experienced person may sometimes be responsible for checking a machine capable of producing extremely polished, confident-looking work.
And polished does not always mean correct.
That is why I don’t think the future can simply be:
AI does the junior work → junior jobs disappear → senior lawyers supervise AI.
There is a missing step.
How does someone become senior?
Perhaps an old idea is coming back: apprenticeship.
When I started thinking about this, I pictured something surprisingly old-fashioned.
An apprentice following a master craftsman.
Watching.
Trying.
Making mistakes.
Getting corrected.
Trying again.
Eventually working independently.
Maybe some professional education will begin to look more like this again.
Except now the apprentice will have AI sitting beside them too.
AI + Apprentice + Expert
AI processes information.
The beginner works with it.
The expert supervises both.
The beginner watches the expert make decisions.
Then they talk about why.
Why did you reject that argument?
Why did you ask that question?
What made you uncomfortable about that contract clause?
Why didn’t you follow the AI recommendation?
What did you notice that I didn’t notice?
That conversation may become more valuable than spending six hours mechanically searching documents.
Stanford Law is already experimenting with something close to this.
This surprised me.
Stanford Law School says it has been adapting its curriculum as AI takes over some work traditionally given to junior associates.
First-year students still learn foundational legal research and writing without relying on AI.
Later, AI is introduced into simulated legal work.
Students may use it to identify legal issues, summarize discovery materials, draft research emails and edit documents.
But there is another fascinating exercise.
Professor Mark Lemley gives students questions accompanied by AI-generated answers and asks the students to grade the AI.
In other words, the assignment isn’t simply:
Find the answer.
It becomes:
Tell me what is wrong with this answer.
Is the citation correct?
What did AI miss?
Is the argument incomplete?
Could there be another interpretation?
Stanford also uses clinical education where students work in a law-firm-like environment with real clients under close faculty supervision.
That looks remarkably similar to a modern apprenticeship.
Learn the foundation → use AI → question AI → practice → receive expert supervision → develop judgment.
Source:
Stanford University — Inside Stanford Law School’s Evolving AI Curriculum
But AI isn’t changing only the lawyer’s side.
This was the second thing that surprised me.
It is changing the client’s side too.
Imagine that I have a legal problem.
I am not a lawyer.
Ten years ago I might have received a complicated contract or court document and understood almost nothing.
Now an ordinary person can ask an AI tool:
“Explain this document in plain English.”
“What are the important clauses?”
“What questions should I ask a lawyer?”
“Organize these events into a timeline.”
“Explain what this legal term means.”
“Help me make a list of documents I should gather.”
That doesn’t suddenly make me a lawyer.
But it makes me a much more informed client.
And that alone could change the relationship between lawyers and ordinary people.
What if people begin going to court with AI?
Then I discovered something even more interesting.
People already represent themselves in American courts.
They are generally called pro se or self-represented litigants.
The U.S. District Court for the District of Arizona provides an entire handbook for people representing themselves.
And its 2026 edition now contains a whole chapter titled:
“Understanding the Use of Generative Artificial Intelligence (AI) in Legal Proceedings.”
The court explains that AI may help self-represented people understand complicated legal concepts, improve writing and formatting, explore potentially relevant case law and understand documents.
But the court gives a very important warning:
The person filing the document remains responsible for its accuracy.
AI can invent cases.
It can misunderstand facts.
It can confidently give an answer that looks convincing but is wrong.
Source:
U.S. District Court for Arizona — Self-Represented Litigant Handbook
Arizona’s Court of Appeals addressed this problem in 2026 as well.
The court noted that self-represented litigants can now ask generative AI to produce complaints, motions and briefs that may look like competent legal work.
But if AI-generated material is submitted without carefully checking the citations and claims, serious problems can result.
Source:
Arizona Court of Appeals — 2026 Opinion Discussing Generative AI in Legal Filings
So the future I was imagining isn’t entirely science fiction.
The courts are already thinking about it.
Maybe the future lawyer will not always do everything for us.
Perhaps ordinary people will increasingly do the first layer themselves.
AI might help us:
understand the problem
→ organize information
→ read documents
→ identify questions
→ prepare timelines
→ understand terminology
Then a professional might enter where judgment, strategy, negotiation, ethics and accountability matter most.
That could also create more interest in limited or targeted professional help rather than paying a lawyer to perform every single step.
But there is an important boundary.
AI-generated information is not automatically reliable legal advice, and representing another person or giving individualized legal advice is regulated. A person should not assume that an AI-generated strategy is equivalent to having a qualified attorney responsible for the case.
So I don’t see this as:
AI replaces lawyers.
I see something more complicated happening.
And then I came back to my child.
This is where the story became personal for me.
As parents, we have spent years asking children questions like:
What do you want to be?
Doctor?
Lawyer?
Engineer?
Accountant?
Designer?
Teacher?
And lately another question has been added:
Will AI replace that job?
But maybe that is not the most useful question anymore.
Because AI may not simply erase entire professions.
It may rearrange how people become professionals.
That means when our children eventually choose colleges, majors, internships and first jobs, perhaps parents should begin looking for something we rarely asked about before:
How does this place train beginners?
Will my child work beside experienced people?
Will someone explain why a decision was made?
Will they see real problems?
Will they be allowed to make supervised mistakes?
Will they learn the fundamentals before relying heavily on AI?
Will they learn how to challenge AI rather than simply accept its answer?
Will someone gradually give them responsibility as their judgment develops?
Those questions may become as important as salary or job title.
Maybe the best first job won’t be the company with the best AI.
Maybe it will be the place with the best human teachers.
A company can give a young person every AI tool available.
But if no experienced person has time to say,
“Come sit with me. Let me show you why I made this decision,”
something important may be lost.
Perhaps shadowing, mentorship, clinical practice and apprenticeship will become more valuable—not less—in an AI world.
That would be an interesting twist.
We spent decades trying to automate away apprenticeship.
Technology may eventually remind us why we needed it.
What I want to remember as a mom
I don’t think I need to predict exactly which profession will survive AI.
No one really knows.
Instead, perhaps I can help my child develop the abilities that sit above the answer:
curiosity,
critical thinking,
communication,
discernment,
responsibility,
the courage to question an answer,
and the humility to learn from someone who knows more.
And someday, when we look at a college program, internship or first employer, I want to remember to ask one more question:
“Who will teach you how to become good at this?”
Because AI may make answers extraordinarily cheap.
But becoming the kind of person who knows which answer to trust, when to question it, and what to do next may take years.
AI may shorten the road to an answer.
But experience is still the road to judgment.
-HappyKay
Editorial Note
This post reflects my personal learning and reflections as a mother thinking about AI, education and the future of work. I use AI as a creative tool to help research questions, organize my thoughts, refine my English and create some illustrations. Because technology and professional rules continue to change quickly, I encourage readers to visit the original sources linked above and make their own informed judgments.
Comments
Post a Comment