DeepSeek, Frederick the Great & The Art of Chasidic Warfare
Tech Tribe Dispatch #285
💡 Think:
In the depths of the frigid winter of 1812, Rabbi Shneur Zalman of Liadi, the founder of the Chabad movement, found himself, his family and some of his followers some 600km from home.
Fleeing Napoleon and his army, whom Rabbi Shneur Zalman viewed as a spiritual adversary to the Jews, the Chasidic leader took shelter in the small village of Pyena, near Kursk.
At 68 year, the journey away from the front had been daunting, and suddenly Rabbi Shneur Zalman took ill.
Days before his passing, he called over his grandson and future successor, Rabbi Menachem Mendel of Lubavitch, wishing to share a memory from the spring of 1764.
That sprint, while visiting his mentor, Rabbi Dovber of Mezritch, Rabbi Shneur Zalman begun to study with Avraham der Malach, “the angel”, the son of Rabbi Dovber.
This week Deepseek has dominated tech industry news - the emergence of a Chinese AI model that competes, perhaps surpasses its American counterparts, at a fraction of with price, has thrown markets and tech pundits into a frenzy.
Begun, the AI Wars have.
In exploring what makes Deepseek’s R1 so unique, I found this summary in the WSJ particularly illuminating:
DeepSeek’s tactic was to cut down on the data processing needed to train the models, using some inventions of its own and techniques adopted by similarly constrained Chinese AI companies.
Imagine the earlier versions of ChatGPT as a librarian who has read all the books in the library, said Lennart Heim, who researches AI at the think tank Rand. When asked a question, it gives an answer based on the many books it has read.
This process is time-consuming and expensive. It takes electricity-hungry computer chips to read those books.
DeepSeek took another approach. Its librarian hasn’t read all the books but is trained to hunt out the right book for the answer after it is asked a question.
Layered on top of that is another technique, called “mixture of experts.” Rather than trying to find a librarian who can master questions on any topic, DeepSeek and some other AI developers do something akin to delegating questions to a roster of experts in specific fields, such as fiction, periodicals and cooking. Each expert needs less training, easing the demand on chips to do everything at once.
DeepSeek’s approach requires less time and power before the question is asked, but uses more time and power while answering. All things considered, Heim said, DeepSeek’s shortcuts help it train AI at a fraction of the cost of competing models.
These three strategies, isolating the specific knowledge domain needed for each query, deploying specialized "expert" models precisely where needed and concentrating computational resources on that specific challenge felt rife for lessons in not just machine learning, but in our own ability to confront challenges…
On strategy for confronting a challenge:
The story Rabbi Shneur Zalman shared with his grandson concerned a spiritual lesson his friend R’ Avraham had shared with them.
The year before the cataclysmic Seven Years war had rocked Europe. During the Battle of Leuthen, fought earlier on in the war, on the 5th of December, 1757, Frederick the Great's Prussian Army beat the larger opposing Austrian forces by outflanking and outmaneuvering them.
There was a lesson to be learned here, R’ Avaraham taught: Within each of us there are spiritual attractions we have to the Divine - the love, awe and mercy we use in the service of our Creator.
Then there is the pull to the earthly base loves, fears and cynicism within us.
Two opposing forces, evenly matched with the full arsenal of emotional and spiritual expression.
The key to addressing the inner challenges, is to outflank the animalistic urges. Like the troops of Fredrick the Great surrounding and destroying each larger, individual flank of the Austrian army, R’ Avraham der Malach taught that we can use all three of our spiritual drives to isolate a single negative trait. Once isolated, we can refine and uplift it.
The lesson here, be it through Deepseek or Fredrick the Great, is one:
It’s a demonstration of how the same fundamental principles of efficiency and strategy can be deployed in confronting problems.
Rather than brute-forcing it, deploying large amounts of resource to confront issues head on, when we isolate the problems we need to confront, focus a given set of ‘spiritual expertise’ we have on an individual issue, we can overcome it.
There’s an interesting detail as well:
Part of the appeal of Deepseek is that the model shows an internal ‘debate’ of different perspectives within itself before giving a final answer… This documentation of the reasoning process feels deeply familiar in the Chabad context.
Just as the Talmud doesn’t just state final conclusions, but instead documents the minority opinions and conversations that led to the final halacha, the Chabad model of Chasidic thought, the one developed by Rabbi Shneur Zalmaan of Liadi, puts strong emphasis on having a deep intellectual appreciation of the divine - not merely the final conclusion, but rather a complex and detailed understanding… only then, when the mind can be fully and deeply engaged, can there be an abiding and truly lasting rest.
🏃 DO:
✨ Join us Friday, February 7th, for our February First Friday curated conversation on Minyan and the Metaverse!
🥂 Don’t miss A Taste of Tu B'Shvat. A bespoke evening of food, fruit, music and conversation for Jews in tech on Tuesday, February 11th!
🌳 Join Tech Tribe for a curated conversation:
Tu BiShvat And the Talmud on Wed, Feb 12, 3pm at Google NYC.
🎩 It’s a family celebration - and you’re invited! Join our son Mendel’s Bar Mitzvah 7pm, Tuesday, February 18th!
🤠 Tech Tribe brings the Jewish vibe to SXSW! Join Us for #openShabbat at SXSW!
🎉 Mazel:
Ben and Karen Lang on the birth of their daughter!
Mazel Adam Lewkovitz on the birth of his daughter!
🌱 #ChaiSociety
Special thanks to the #ChaiSociety members who donate monthly to support our community!
Adam Kotok, Avrohom Margolis, Bill Dilworth, David Lobron, Cal Nathan, Chanoch Rosenberg, Fitz Tepper, Hershey Gutnik, Jacob Ner-David, Lev Podelko, Levi Plotkin, Liel Leibovitz, M. Kahn, Manilan Houle, Mattanya Ginsburg, Melissa Davis, Mendel Konikov, Michael Morgenstern, Nikita Sokolsky, Scott Shapiro, Sholom Laine, Tzali Reicher
Tech Tribe is entirely community-funded - Invest in Tech Tribe (starting at $36 a month), and join the monthly #ChaiSociety!
📚 Read:
🕴🏻 Max Chiswick, 39, Adventurer, Innovator and Man of Faith
🏡 In Lubavitch a Home Was Built The Life and Times of Rebbetzin Rivkah Schneersohn.
🔥 Lit:
This week, light Shabbat candles in NYC at 4:55 PM
For Shabbat candle-lighting time in your area click here.
Like what you saw? Want more? Please feel free to forward this email far and wide!
Support Tech Tribe’s mission to build community and curate epic experiences for young Jews in tech and digital media!