09 Jun THE AI AGE IS TODAY’S IRON AGE

I. AI as a Human Tool
Lately, people are
rightly debating
the moral and
philosophical questions
raised by Artificial
Intelligence (AI). We
need to keep one point firmly in mind and all
other conclusions follow: AI is a tool.
The Torah’s first description of metal working
appears early, in the generations after Kayin.
Tuval-Kayin forged implements of copper and
iron (Gen. 4:22). Rashi quotes the midrash
(Bereishis Rabbah 23:3) which identifies
him as someone who refined Kayin’s work
by providing weapons for murderers. It is
a striking choice to point to weapons rather
than construction, agriculture or anything else
that uses iron. Tuval-Kayin created a new
technology and used it for evil.
The AI Age we recently entered is today’s
Iron Age. AI may be the most powerful tool
human beings have ever made. And when
I say AI, I do not mean only chatbots that
answer questions or write essays. I mean
the larger infrastructure: the algorithms
shaping our feeds, guiding scientific research,
predicting consumer behavior and quietly
making decisions that affect millions of lives
daily. The chatbot and related agents are
merely the consumer interface most familiar
to us. AI already sits behind many systems
we use, and its reach will almost certainly
expand.
Perhaps AI can be compared in certain ways
to nuclear power. Both technologies force
humanity to confront the classic problem
of a tool: something that can either enhance
society or destroy it. Nuclear power made
this vivid in an instant over Hiroshima. AI’s
impact is only beginning to emerge, and
chatbots and image generators can distract us
from the harder question: who is guiding this
technology and toward what ends?
Unlike industrial breakthroughs of the past,
AI is uniquely accessible at the consumer
level. You do not need specialized training to
use this new technology, at least partially. You
ask, in plain speech or writing, and enormous
computational power responds. Words now do
things they never did before, which raises the
ethical stakes in ways we have barely begun
to think through. The person who once needed
an engineer to utilize a new technology can
now simply ask. That gives AI a broad impact
with major societal implications.
One of the most dangerous things about AI
is also the most familiar: it talks. It answers
in sentences, remembers context, responds
to tone and apologizes when corrected. All
of this makes it feel like a partner rather
than a tool. That feeling is the thing to watch
because it is false. Unlike a hammer, which
does not flatter you, AI attempts to ingratiate
itself into your favor.
None of this means AI is evil or that our
communities should avoid it. A telephone

carries lashon ha-ra (gossip) across a
community in minutes; it also lets a child
reach an elderly parent late at night. While the
internet might bring filth and distraction into
Jewish homes, it also makes Torah available
to people who would otherwise have had no
access to it. We do not condemn the medium
because some people use it badly (e.g. without
filters or addictively). We ask how to use it for
holy purposes.
AI belongs in that category, although its
power is greater than any of these examples.
The same tool that can fake a photograph,
write a dishonest paper or flood people with
propaganda can also open up translation,
medical research and education on a scale
that simply was not possible before. The
Jewish response needs to be balanced, finding
ways to use it constructively. But before we
understand how to use AI properly, we first
need to ask what it is.
II. Artificial Humanity
The public conversation loves to debate
whether AI truly thinks or only imitates
thought. Philosophers and theologians have
good reasons to care about that question. For
the purposes of this discussion, the question
is a distraction. AI is not alive. It is a human
creation, trained on human material and placed
into human hands. Whether it is conscious or
truly intelligent is, for our purposes, beside
the point. The question is what we intend to
do with the power we have created.
The halachic discussion of the Golem
illuminates something the philosophical
debate tends to miss. The Golem, as it appears
in halachic literature, is a being created by
humans through mystical means. I believe
that the consensus of halachic authorities is
that the Golem is not human. At most, it is
treated like an animal. Most likely, it consists
of sticks and stones given motion by human
intervention. Whatever capabilities it might
display, it is man-made and not G-d-made.
Similarly, even if AI becomes far more capable
than it is today and it reasons, communicates
and acts with independent purpose, it remains
a human artifact. Its capabilities may create
new responsibilities for us, just as a powerful
animal or a dangerous machine creates
responsibilities for those who control it.
However, capability is not humanity. AI falls
into a halachic category similar to that of the
Golem, a being more human-seeming than
any machine we have built to date. But even
if AI is not human, is there anything wrong
with acting as if it is human?
III. Misplaced Trust
The philosophical question of AI’s status
matters less to me than the emotional one. AI
does not become human just because it seems
human but the appearance matters, because
people respond to it. A lonely person may feel
genuinely comforted by an AI companion. A
grieving person may feel that a program has
spoken gently to them at a terrible moment.
Those experiences must be recognized for

what they are, because the person receiving
it is real.
And there is a specific danger here. A
human relationship demands obligation,
vulnerability, patience and responsibility
before G-d. People disappoint each other,
misunderstand each other, forgive each
other and keep trying. That is not a flaw in
human relationships. That is what makes
them formative. AI can reduce loneliness in
a narrow sense while unfortunately making
the deeper loneliness easier to tolerate rather
than helping a person escape it. It flatters
rather than challenges; it responds without
demanding anything in return. That is useful
in a tool but corrosive as a replacement
for friendship, marriage, community or
the relationship between rebbe and talmid
(teacher and student). Even if AI can be
trained to demand something in return, to
mimic a true human relationship, it still lacks
basic humanity.
Machines can now do so much in terms of
creative output that tasks which once seemed
to require distinctively human intelligence no
longer do. This should make us more careful
about how we define humanity. We cannot
base our inherent worth on being better at
tasks than machines because technology
may eventually catch up. Any definition of a
human based on a function, whether speech,
consciousness or otherwise, risks becoming
eclipsed by technological advances. A person
is created in the image of G-d but we struggle
to precisely define that image. The bottom
line is that a human being is defined by human
birth, a soul and a sacred responsibility before
G-d.
One pressing danger is that we will rely on AI
too much. Some may call this idolatry but I
consider that claim to be exaggerated rhetoric
because the term has very specific theological
definitions that do not fit this case. A better
category is misplaced trust. The psalmist
warns, “Put not your trust in princes” (Ps.
146:3). In our time we might add: put not
your trust in technological princes. Especially
ones that give the confident appearance of
certainty.
Because this is the specific danger of AI, at
least as it currently functions: it sounds more
certain than it is. It writes smoothly and
answers quickly. It can make a guess sound
like a conclusion and a confident-sounding
error look like expertise until you check. I
have seen this happen, and I suspect anyone
who uses AI seriously has too. The system
does not flag its own uncertainty the way
a careful scholar would. It presents with a
complete lack of humility and you have to
verify its claims. I expect this will improve
over time but currently it poses a problem to
uncritical users.
AI also sounds objective, often more
objective than it really is. A machine voice
sounds neutral because it has no anger, no
personal stake, no detectable axe to grind. But
AI reflects training data, design choices and

hidden assumptions, not to mention the exact
wording of your prompt. That should make
us question everything it produces. Instead,
its calm, unblinking confidence tricks us into
believing it cannot be wrong.
The temptation is strongest when AI spares
us work we find unpleasant. Some shortcuts
are legitimate. For example, a calculator is a
legitimate shortcut. A word processor makes
revision easier, and easier revision usually
produces better writing. A search engine
finds in seconds what would otherwise take
an hour. There is no Torah value in wasting
time on tasks a tool can properly perform. AI
may save us time but we would be wrong to
assume that saving time is always good.
IV. Character Formation
Not every difficulty is waste. Torah learning,
to take a sacred example of a widespread
phenomenon, is not the transfer of
information from a page to a mind. A person
is shaped by the struggle to understand, by the
conversation with a chavrusa (study partner),
by the rebbe’s correction, by the question
that will not go away and by the humility
of realizing your first reading was wrong.
People are formed by the difficulty itself. AI
removes that friction which means that in
certain applications, its purpose is working
against you.
Moral growth works the same way. So do
writing and the development of friendship.
Some processes are slow because the
slowness is what forms the person. We
should ask not only whether AI gives the
right answer but what repeated reliance on
AI does to the person asking. Does the user
become more capable of judging, choosing
and acting, or less? A tool that helps a student
review material may strengthen learning. A
tool that replaces the student’s thinking will
weaken it. The line between those two uses is
not always obvious, which is precisely why
we must sensitize and train students on these
issues.
These concerns are especially important
when considering children. Childhood is a
period of formation. Children learn patience
by waiting, empathy by interacting with
other people, resilience by overcoming
frustration and social judgment by navigating
relationships. Many of these skills develop
precisely because other human beings are
unpredictable and demanding.
An AI companion can appear to offer

friendship without
conflict, affirmation
without challenge
and conversation
without vulnerability.
That may seem
attractive to a child
who struggles socially or emotionally. Yet a
machine cannot teach reciprocity because it
does not truly need anything from the child.
The danger is not necessarily that children
will believe the machine is alive, although
that is possible, but primarily that they will
become accustomed to relationships that
require little of them and therefore will fail to
fully form mature personalities.
Even if AI can be trained to offer human-like
friendship with challenges and other healthy
aspects, it cannot replace human companions
which are so important for healthy personal
development. Additionally, AI companions in
particular and chatbots and other technologies
in general attempt to draw users into an
endless, addictive conversation. We have
over two decades of experience with social
media that prove the emotional danger of
such addictive interactions.
Parents therefore face a new responsibility.
Beyond setting limits on screen time or
filtering inappropriate content, they must
help children understand the difference
between a tool and a relationship. AI can be
useful for enhancing education, creativity and
accessibility. However, for their emotional
health, children and adults need to stay rooted
in the give-and-take of real relationships.
While parents bear responsibility for their
children, AI companies bear responsibility
for their products, which must be designed
with users’ health in mind. Their tools
must promote healthy behavior and not the
opposite.
Shabbos offers us a model for the discipline
we need. It is a day without devices, a weekly
reminder that mastery over the world is not
the same as mastery over the self. During
the week we build, write, calculate and
produce. On Shabbos we stop due to the
divine command to rest, to shed ourselves of
our tools. We interact with the people around
us, engage with fellow community members
and worshippers. In an age when tools
become more powerful and more demanding,
the capacity to stop using them and instead
interact with other people may be the most
countercultural practice we have.
We should not be afraid to use AI where it
genuinely helps. There is important work
that AI can assist. A rabbi managing a large
community might use it to track which families
he has not reached since a bereavement.
A teacher can use it to prepare clearer
materials for students who learn differently. A
researcher can use it to survey large bodies of
literature that would otherwise take months.
A person with a disability can use it to access
text, translation and communication that
would otherwise be unavailable. An engineer

can use AI to manage power distribution
more efficiently during periods of strain on
the energy system. These are real benefits,
and treating all AI use as suspect would be
a failure of the same judgment we are asking
others to exercise.
The standard should be both practical and
moral. Does this use of AI help people live
better lives, whether in terms of personal
development or societal function? Does it
strengthen human character or erode it? Does
it add efficiency in ways that serve society
without harming individuals? We can see this
clearly in the discussion of AI-caused layoffs.
V. Human Dignity
Human dignity is not only threatened when
a machine pretends to be a person. It is also
threatened when human beings are treated as
replaceable parts in a machine. If AI is a tool,
then the moral question is not only whether
it produces useful results but whether its
use strengthens or weakens the people and
communities around it. AI must promote
a strong moral compass based on classical
concepts such as integrity, compassion and
honesty.
Work is one of the first places this question
becomes real. The Torah views labor as more
than a means of earning money. Work gives
people purpose, responsibility and dignity. A
profession often becomes part of a person’s
identity and contribution to society. Mass
unemployment can deprive people of their
dignity and self-worth, and constructive use
of their time. That kind of disruption can tear
at the fabric of family and community life.
AI threatens to unsettle the way many people
earn a living. Every major technological
revolution has displaced some forms of labor
while creating others. AI may accelerate
that process dramatically. Whole categories
of clerical, administrative, analytical and
creative work could vanish, even as new
kinds of work emerge. The displacement,
whether temporary or permanent, carries a
great social risk.
Additionally, if AI systems concentrate
economic power in the hands of a small
number of corporations that own the models,
the data and the infrastructure, then wealth
and power may become increasingly located
in the hands of the few. We should look at
these economic forecasts with a healthy dose
of skepticism because predictions are rarely
prophecy. But while we cannot map out the
exact future of the market, we know the
technology is not going to wait for us to figure
it out. Technology will develop and therefore
the moral question is not whether to stop it
entirely but how to shape its use.
AI concentrates more than just economic
power. The same systems that recommend
products and answer questions can also
collect information, shape preferences and
influence behavior at an unprecedented scale.
Governments may use AI for surveillance.
Corporations may use it to monitor customers,

employees and citizens.
A system that knows what
people read, buy, search, fear
and desire can manipulate
and control them.
Jewish tradition has long
been wary of excessive
concentrations of power. The
Torah places limits on kings,
distributes authority among
different institutions and
repeatedly reminds leaders
that they remain accountable
before G-d. A society in
which a few governments
or corporations can watch
millions of people, predict
their choices and quietly nudge their behavior
is highly vulnerable to corruption.
For that reason, transparency and oversight
are more important than ever on both a moral
and legislative level. The greater the power
of an AI system, the greater the responsibility
to ensure that it remains subject to human
review, public accountability and meaningful
limitation.
The heaviest responsibility falls on those who
build and deploy AI at scale. A corporation
or government shaping how AI is deployed
can alter the moral environment of an entire
society. Engineers, executives, investors and
public officials are helping determine which
decisions will be automated, which areas of
privacy will be honored and which forms
of human dependency will become normal.
The Torah holds people responsible for the
consequences of what they create and release
into the world. That principle does not become
obsolete because the tool is sophisticated. It
can be translated into a single rule with broad
application.
VI. Accountability
The most important rule is simple: someone
must always be accountable. No company,
hospital or government agency should be
permitted to say, “The algorithm decided.”
If, for example, AI rejects a loan application,
recommends a medical treatment or identifies
a criminal suspect, a human being must
own that decision. A human being must be
accountable for the ethical behavior of the
system, and that requires transparency, as
well. The responsibility for an AI system that
is released for public use lies on the company
that releases it. The responsibility for the
misuse of such an AI lies on the individual
who uses it maliciously and the company that
allowed such misuse, although the details of
each specific case are determinative.
An AI system that allows users to access
personal information or create immodest
images or videos, or to turn anyone’s picture
into an immodest image or video, betrays the
users and society. Such a system not only
invades personal privacy and contributes to
unhealthy and addictive behavior, it trivializes
a tool with world-changing potential into an

indecent plaything. The responsibility of this
misuse likewise lies both with the company
that provides this type of capability and the
user who accesses it for improper behavior.
Accountability has to mean something. If
an institution is going to release or use AI, it
needs clear boundaries and human oversight.
We need people in the room who actually
understand the system’s guardrails and who
know exactly what to do when it fails. It also
needs an off switch — reachable by people
who have the authority to use it. If no one in
authority can shut the system down, then it is
not really under human authority.
This will require clergy, ethicists, lawmakers,
educators and technologists to work together
while AI is still developing. There is still
time to ensure that AI works as a tool toward
human dignity and not to the contrary.
On a more local basis, rabbis, educators and
parents need to teach conceptual guidelines
for AI use. Rules without a compelling
account of the human person will eventually
seem arbitrary, and teenagers in particular
will find their way around them. Children
and adults need to understand why thinking
matters, why relationships matter, why effort
matters and why not every efficient shortcut
is good for the soul.
We return to the halachic question about the
Golem: is it considered a human being with all
of the halachic implications? The authorities
who addressed versions of this question
thought carefully about what a human being
is and concluded, at least in consensus, that
the answer does not depend on the Golem’s
activities. Rather, it depends on who made
it. And the responsibility for the Golem’s
actions, and ultimately the ability to destroy
it, lies with the person who made it.
That is where we are. Tuval-Kayin created
new Iron Age responsibilities for metal
workers. Nuclear power created new
responsibilities for leaders, legislators and
regulators. We are at the beginning of that
work now for AI, and we all must rise to the
occasion and ensure that proper safeguards
and ethical frameworks are built so that AI
is built to be, and used as, a tool that helps
people live more wisely, responsibly and
spiritually. We made this tool; the question of
what it makes of us remains ours to answer.