Act before GenAI becomes anti-Zionist

Why everyone who cares about countering anti-Zionism needs to act now to influence the field of artificial intelligence
The TikTok battle flaring in the United States has particular resonance to the Jews. Since October 7th, TikTok and other social media tools have been awash with anti-Zionist and antisemitic content, convincing young adults across the US of the injustice of the existence of the Jewish state. Even before the war, studies showed that the Jewish state was the most attacked country in the world on social networks. Anyone who spent time on campuses in the past two decades can tell you that Israel ranks below North Korea as a subject of scorn.
Yet, as a friend working in a large company building AI tools shared with me, there is a difference between a group of students marching down the quad shouting “from the river to the sea” and a digital tutor running an artificially intelligent anti-Zionist chatbot. “People don’t understand how bad it is going to get,” he told me. “The people writing the rules on the ethics of the models aren’t Israel’s biggest fans. You’ll see that soon enough when your kids do their homework about Israel and the chatbot starts telling them that Israel is a colonialist entity.”
As the ex-Googler Shaun Maguire told the Free Press, Google’s Gemini AI’s ridiculous portrayal of Black Nazis and Asian Knights “was not a one-off incident. It was a symptom of a larger cultural phenomenon that has been taking over the company for years.” These overt biases featured by the press are just the tip of the iceberg. A study published this month exposed how covert bias baked into algorithms can be even more pernicious than open hate speech. These biases were not, as far as we know, intentionally implanted; they are the result of the content used to train the large language models (LLMs) at the heart of Generative Artificial Intelligence (GenAI) such as Gemini and ChatGPT, and the blindness of moderators to the bias they may unknowingly share.
We need to act now because lock-in will set in soon, creating a path of dependency that will be nearly impossible to break. Like VHS vs. Beta, Netflix vs. Blockbuster, Google vs. Altavista, or Facebook vs. MySpace, it is most likely that there will be one clear market leader in GenAI. Since the people overseeing the culturally-attuned dimensions of the models have all themselves been trained in the world’s leading academic institutions, it stands to reason the model they oversee will be biased according to the biases propagated by these institutions: the anti-Zionist worldview denying the Jewish people what it grants every other people. While the employees of market leading firms developing GenAI models may be well meaning, we cannot afford to leave our destiny in their hands. Especially since we are already seeing existing models used in troubling ways concerning Israel.
We have to act now because any later will be too late. There is a reason the same Saudi Arabia which has poured billions into the world’s elite institutions of higher education has just recently established an International Center for Artificial Intelligence Research and Ethics in Riyadh. A reason why Iran is investing so heavily in developing algorithmic training infrastructure and content creation capabilities. A reason why Hamas’ allies are spending so heavily on the social media war, generating mountains of anti-Zionist content to influence the training of future models to tip the scales in favor of their conception of ethics, an ethics that excludes Israel. Our enemies are strategic. They recognize that no tool is better suited to shift the tide of global consciousness than GenAI.
We cannot rely on the Israeli government to act. It is, frankly, too unstable. Even taking into account the allegedly extraordinary efforts carried out by Israel’s covert services, the nature of this industrial battle requires a sustained public engagement with the market leaders. Engagement only civil society and business leaders can reliably sustain.
To win the war, Israel and Jews committed to a Jewish state are in urgent need of a coalition to finance and build a transdisciplinary institute to leapfrog Israel’s position in GenAI. This coalition should have a particular focus on the use of GenAI in education, and ensure it is founded on the ethics Rabbi Jonathan Sacks called the “dignity of difference.” We need to immediately attract at least a dozen of the world’s best researchers and visionaries, experts in ethics and editorial, many of whom happen to be Jewish. The price tag will be big – in the billions – but no amount of future defense funding will be able to prevent the catastrophe waiting to happen if we do not make this investment now.
The priority of such an institute should be to embed the ethics of pluralism into the foundations of AI, to ensure it reflects the right of all people, even and especially Jews, to self-determination. It should create open-source, publicly scrutinized tools to help communities monitor and enforce this pluralist ethics. It will require writers, poets, political scientists, marketers, behavioral economists, sociologists, and experts in international affairs to conduct the political and public diplomacy required to build coalitions among the nation states and the digital empires to craft regulation that enforces this ethic. Only by taking bold action at this particular moment in time will we be able to repel the anti-Zionist takeover attempt underway.
White the godparents of GenAI may be Jews, it is GenAI that poses the greatest potential to turn humanity against our right to self-determination. We cannot allow that to be our fate. It is time for philanthropists and technologists to found an emergency initiative on the scale led by Oppenheimer to build models and coalitions and standards that are aligned with our values, values that will benefit all peoples and their particularities. To encode the dignity of difference into the digital architecture of our society. The natural place for such a project should be Israel, and the only time to start it is now.