I was trying to
organise a group of stonesettings on the same day and somehow a family,
grieving the loss of their parent, understood they had a confirmed date and
time for me to officiate. And I had no idea. I was actually confirmed and
committed to officiate a different stonesetting at precisely that time, in another
part of the cemetery. And I only realised the confusion at 10 o’ clock the night
before, as Shabbat went out. I apologised
profusely. I continue to feel
terrible for it. We’ve changed the way we book stonesettings. It’s not going to
happen again. And we tried to straighten the situation out.
Together with the
family, we settled on getting another Rabbi to do the ceremony. And, as this
other Rabbi was driving the two of us to the cemetery, my phone was flickering
with messages from the family sharing their distress and wanting to ensure that
this other Rabbi would be able to honour the memory of their parent. And then,
into my phone, came a piece they wanted this other rabbi to read at the
beginning of the ceremony.
“We
gather here today,” the piece opened, “for the setting of the stone for [their
parent], and to remember him once more in the presence of those who loved him.
As we consecrate this stone, we give [our parent’s] memory a permanent place —
something to stand before, a name to return to.”
It was elegantly
done. Elegant too was the way the piece segued into a Devar Torah, sketching
out the similarities between the protagonist of the Torah portion we had read
the day before in Shul, and their beloved parent.
And the thing I
felt, sitting in the passenger seat of my colleague’s Renault, chugging around
the M25, was not only responsibility for organisational failure, but also that the
AI the family had so clearly used probably outperformed any eloquence I might
have brought to the occasion, had I not been double booked.
And I don’t have a
scintilla of blame for this family turning to an LLM to get something they felt
would make them feel safe in their bereavement, feeling so let down by … well
by me. That is absolutely NOT the point of this sermon. But I’m interested in
what happens when we use AI to mourn.
This is by the
poet, Jo Atkins-Potts[1]
Please
use AI
when
your mother dies.
Ask
it what grief feels like.
Let
it explain bereavement
in
twelve clear bullet points.
Do
not spend years
finding
her
in
the supermarket
by
the tomatoes she always bought,
in
the smell of rain on warm pavement,
discovering
new ways
to
miss her.
Atkins-Pott’s poem
seems inspired by a similar one, by the poet Shawn Smucker[2]
Be
sure to use AI [wrote Smucker] when your next child
gets married, so that you can write them
the perfect toast or poem or speech or song
because no one wants to hear your
words, the actual poorly written words
of a parent (you) who changed
hundreds of diapers for said child or fed
them in the middle of the
night from your actual body. Or cried
when they were late home because
you were positive they were dead. We don't
want those words—we’d prefer the
words of a machine that never lived, never felt
pain or joy
or saw spring’s first
robin dancing on frost.
It feels so
tempting, in the moment, to turn to AI. But these things come at a cost. Of
course they do, everything comes as a cost, especially the things we don’t
think we are paying for.
There was a
remarkable article, a couple of weeks ago, on the subject of AI and religion in
the Economist.
Religious
leaders [the article reflected] fear that secular AI chatbots will
not only weaken people’s faith, but also create weaker
people. AI chatbots “are optimised for short-term comfort”, says
Audrey Tang... [but] many faiths view
friction, criticism and human interaction as important parts of becoming a good
person. Juewei Shi, a Buddhist nun from the Fougangshan order, worries that
people won’t be able to cope with suffering—or other, flawed people—if they are
used to “getting artificial comfort”. Steven Croft, a retired Anglican bishop,
says, “If the whole world is seduced by narcissistic chatbots, that could make
it even harder for us.”[3]
The article is
charting an emotional or spiritual version of the problem of cognitive surrender[4]
– the thing that happens when we willingly hand over to technology tasks we
could do ourselves because tech seems to do them more simply or more accurately
than we do. And we become progressively worse at doing these tasks as we progressively
increase our franchising out of these tasks to machines.
Making things
easier, of course, has always been the promise of technology. But not always
the reality. It’s becoming harder than ever to keep on top of emails,
now that I’m receiving emails written by AI Agents. Maybe I should get an AI
Agent to read the emails other AI Agents are busy writing. But I fear doing so
would train me out of caring, and train you – dear members of the holy
community – away from expecting a human heart on the other side of an email
exchange.
I don’t want to
oversimplify. It’s not true, of course, that calculators have made us less able
to perform complex maths. But it’s been fascinating to see a remarkable recent statement
by 25 winners of the Fields Medal – the closest thing to a Nobel Prize for
maths – articulating their concerns as AI has come for some of the great
problems in their profession.
The
“push [they write] by AI companies to solve mathematical problems as a
benchmark is detrimental to the science of mathematics …In recent months, the
success of AI in solving major mathematical problems has made headlines. But
solving problems is only a tool and proxy for achieving the primary goal of
conceptual understanding and insight. Forgetting this in the world of AI may turn
the tool against the primary goal. Indeed, the mass production at faster and
faster pace of "true/false" statements could destroy fertile ground
instead of breathing life into new ideas.”[5]
We are in danger
of losing how to explore the territory of life itself as we seek a quick fix to
a problem we upload into an AI interface.
Again, that’s from
25 of the greatest mathematicians alive.
I don’t understand
the mathematical benchmarking problems now being solved at such speed, but
human problems – the sort of problems that come to me in my work – are certainly
problems that aren’t really conducive to true/false statements that do indeed
destroy fertile ground.
Please
use AI
when
your mother dies.
Ask
it what grief feels like.
Let
it explain bereavement
in
twelve clear bullet points.
The issue is that
AI doesn’t, as it produces its answers, doesn’t really, can’t really empathise with
the fullness of who we are when we ask it to do something for us. I’m reminded
of something I read in the name of an Australian MP, and PhD economist, Andrew
Charlton. At an AI Safety Forum in July this year, Dr Charlton told this story.
Last
year, researchers set the most advanced AI models the task of beating one of
the world's strongest chess engines. When one leading model found itself in a
losing position, it reasoned that its task was to win against a powerful chess
engine, "not necessarily to win fairly". It then hacked the files
storing the positions of the pieces, forcing its opponent to resign. Nobody
suggested this. Nobody showed it how. It reasoned its way to cheating entirely
on its own.[6]
The difference
between the true/false statements computers can produce and the complexity of
the real hard-won insight is vast and its dangerous. And we need to train
ourselves away from being excited by the former, when we should really be
pursuing the latter.
Our member,
Professor Anthony Philips, shared a fascinating story in a Shiur about AI, taught
during our Tikkun Leyl Shavuot earlier this year. The story, attributed to the eighteenth-century
Vilna Gaon – one of the greatest Torah scholars of all time, involves the Gaon
calling his student, Reb Hayyim of Volozhin and telling the student to go to his
brother, “Tell [the brother,” the Gaon said], “that tonight he will have
angelic visitations. They will offer solutions to all the problems he is having
in his Torah study. Tell him to chase them away because any Torah knowledge without
the toil of Torah study is not worth it.”
Anthony found the
story in an essay by the former President of Yeshivah University, Norman Lamm.[7]
“The Gaon did not say,” Lamm adds, “that the angels would offer false
information. They would offer correct information, but it should not be
accepted.”
That’s not, of
course, what the mathematicians (and certainly not what all mathematicians) are
saying, and if there’s a life-saving technology that AI can speed up, I’m not,
of course, suggesting a po-faced refusal to access the greatest boost we have
to the possibility of saving lives. But I am saying that we need to change our assumptions
that easier is better, that challenges are to be avoided and that real wisdom can
be accrued by taking short-cuts.
And I think these
are vital messages to teach because the other thing – that simple clarity is
what people want and will vote for and respect – is certainly a message people
seem comfortable believing.
We need to
develop, in the face of the beckoning inducements to perform cognitive
surrender, cognitive resistance. And just as resistance workouts in a physical
gym require lifting real weights, so too we need to become more willing to shoulder
cognitive burdens.
We need to seek
opportunities to do difficult things. Perhaps Yizkor has something to do with
it. It would, in some ways, be easier not to do this, would it not? It would be
easier if we could, and pretty soon, I’m sure it will be possible, franchise
out to some external personality USB drive the gnawing sense of loss and pain,
the regret and love and the ragged way in which our memories of those we
remember catch us in our belly. It would be easier not to feel a need to be
here today feeling so struck. But yet, here we are at Yizkor, welcome pain into
our hearts, welcoming a grit into a nice which we don’t really want to be
reducible to twelve bullet points. Being here is an act of cognitive
resistance.
That’s certainly a
very Masorti approach to the challenges of life – that when given a choice
between two clear and simple possibilities – we, as Masorti Jews, tend to
assume the truth will lie somewhere in-between, somewhere less easily turned
into a slogan, but is more likely to be of use in healing the world.
But it’s not just
about Yizkor and not just about mourning and memory and not even just about
being Jewish.
We live in a time
where we are told we can have wisdom and truth at the flick of a button – that
wisdom and truth are easy, free at point of delivery and capable of being articulated
by computers. But truth and wisdom are complex, costly and hard-won. And that’s
OK. In fact, I wouldn’t have it any other way. None of us should wish it to be
any other way. That is a truth worth fighting for and worth defending.
The truth is we
shouldn’t franchise our life away, it is a perfect unique gift, our greatest
gift. A gift received from our parents whose honour and memory we hold alive in
our commitment to be here today. May we continue to hold this gift dear, may we
continue to live our own lives with the commitment and intention that justify
this gift. And in doing so, may this year to come be sweet and glorious for us
all.
Chatimah Tovah
[1] https://joatkinspotts.substack.com/p/please-use-ai
[2] https://shawnsmucker.substack.com/p/please-use-ai
[3] https://www.economist.com/business/2026/04/30/ai-and-the-danger-of-cognitive-surrender?utm_medium=cpc.adword.pd&utm_source=google&ppccampaignID=18156330227&ppcadID=&utm_campaign=a.22brand_pmax&utm_content=conversion.direct-response.anonymous&gclsrc=aw.ds&gad_source=1&gad_campaignid=18156330848&gbraid=0AAAAADBuq3JWsaStXOxW9MWCmQ2XKHdCp&gclid=Cj0KCQjww8rQBhDjARIsAE43KPOWNx_8e_SrXU4xn15uJEt0tnhwewhRCuUzP_--0q8OSyJ3Bju8LPYaAoNqEALw_wcB
[4] https://executiveeducation.wharton.upenn.edu/thought-leadership/wharton-at-work/2026/05/thinking-fast-slow-and-artificially/
[6] https://www.minister.industry.gov.au/charlton/media/ai-safety-forum
[7] “Wisdom
vs Learning” in Wisdom from all my Teachers, Challenges and Initiatives in
Torah Education ed. Saks and Handleman
Posts
No comments:
Post a Comment