The voice behind the machine: A Palestinian’s AI solution to a long-standing dispute

Adnan Jaber has developed dialogue technology that might accomplish what traditional mediation has failed to do: finding consensus in seemingly irreconcilable political disputes.

Photo courtesy of Adnan Jaber

In an era when dialogue across the Israeli-Palestinian divide has become increasingly rare, a Palestinian technologist in Los Angeles is testing whether artificial intelligence (AI) can succeed where human facilitators have struggled.

Adnan Jaber has spent the past year developing what he calls a “dialogue facilitator”—an AI system designed to mediate conversations between opposing sides in political conflicts, extracting points of agreement and gently steering discussions away from familiar patterns of recrimination.

The technology could represent a new frontier in conflict mediation, one that operates below the level of formal diplomacy but above the chaos of social media discourse. The innovation targets the micro-level interactions that could eventually support broader political solutions—building the connective tissue that formal diplomacy presupposes but rarely supplies.

During my first encounter with Jaber’s technology, I was startled by its simplicity. Melanie Robbins (the deputy director of Realign For Palestine) and I listened as the AI bot—speaking with Jaber’s own voice—mediated a live conversation. “Good point Melanie,” the artificial voice said, before paraphrasing points, flagging areas of agreement, and steering the discussion forward in a way that felt remarkably human.

The technology appears able to mirror skilled facilitation; meanwhile, it is infinitely scalable. Therefore, it could represent a fundamental shift in how consensus-building occurs in deeply divided societies.

Jaber was born and raised in East Jerusalem to an Arab Muslim family, and he told me that in his twenty-two years living there, he never had a single Jewish friend. He said that barriers to dialogue with his Jewish neighbors seemed insurmountable—he didn’t speak Hebrew, they didn’t speak Arabic, and the divide followed him into his professional life. “I would show up for interviews in Tel Aviv, sit at the table across from someone wearing a kippah, and there was always this tension,” he recalled. “Neither side really trusted each other.”

He told me that a program called Tech2Peace changed everything for him. The initiative connects young Palestinians and Israelis through high-tech training and dialogue workshops, and Jaber explained to me that it provided his first real experience with structured conversation across the divide. During empathetic listening sessions where participants shared stories about identity and conflict, one aspect caught his attention above all others: the facilitators themselves. “They’d open hard topics, give people time to speak, summarize for everyone’s understanding, invite contrary opinions, raise or calm the tone depending on how the discussion was going,” he explained. “I really liked how facilitation was done.”

But Jaber added that he also saw constraints, because it’s not realistic to imagine a trained facilitator in every room or in every conversation where such mediation might help. So, he told me, he decided to build a voice that behaves like one and can be scaled infinitely.

The facilitator doesn’t arbitrate who’s right. Instead, it does something simpler and more functional: Every few exchanges, it produces what amounts to a receipt showing what both parties have agreed upon, then asks a follow-up question intended not to enrage, but to bring unavoidable questions to the fore. Though Jaber is characteristically humble about the technology’s limits—he acknowledged that “AI can move minds, but it doesn’t always move hearts”—there’s something unique about the technology’s ability to encourage a fragile consensus and steer a conversation away from polarization.

Jaber’s prototype is unglamorous—no gleaming app or slick interface. Two people speak, and the system listens, paraphrases plainly, flags overlaps, and nudges the conversation forward. It works on a simple phone line, a choice Jaber said he made to improve the tool’s accessibility. The people he wants to reach often do not have smartphones and only have access to inconsistent data (or none at all). In Gaza, Israeli strikes have destroyed telecommunications infrastructure—including fiber cables, transmission towers, and switching stations—and fuel shortages have knocked out backup generators, diving the territory into multiple near-total communication blackouts since October 2023. Palestinian operators report that roughly 70 percent of the network has been damaged or destroyed, leaving much of what remains operating on outdated 2G connectivity.

But accessibility is only half the challenge. Considering reports that Israel has used AI tools to generate targeting lists during the war, many Palestinians would view any technology that records their voices with deep suspicion—no matter who built it or why. The fear that transcripts or even participation data could be intercepted and weaponized isn’t theoretical; it’s rooted in the lived reality of surveillance in conflict zones.

Despite the issues, similar innovations elsewhere have shown promising results. Taiwan has deployed a vTaiwan platform for consulting with the public, which uses tools such as Pol.is for opinion mapping to help generate “rough consensus” on everything from Uber regulations to copyright law. The United Nations’ Innovation Cell has pioneered digital dialogues in conflict zones such as Libya and Yemen, bringing citizen voices into international negotiation circles.

While it may appear disheartening that machines are needed at all to mediate very human conversations, Jaber told me that sometimes humanity needs to accept help wherever it is found. His response captured something essential about the current moment in conflict mediation—the recognition that traditional approaches have reached their limits and that new tools might open possibilities that human facilitators, however skilled, often struggle to access.

For Jaber, these realities make it necessary to have auditable prompts, transparent summaries, and the publication of both agreements and dissent. Currently based in Los Angeles after wrapping up a teaching semester at the University of California, Los Angeles, he’s preparing to register a company for Jaber AI while grappling with the harder questions his work raises.

Yet, even the most sophisticated dialogue technology means little without uptake from decision makers. Taiwan’s vTaiwan succeeded partly because of visible government integration. For initiatives like Jaber AI to achieve lasting impact, they must feed real decisions rather than merely populate dashboards with interesting data.

Jaber hasn’t yet tested his facilitator with Palestinians and Israelis on the ground, having only had opportunities to probe it with diasporans thus far. When I suggested integrating the facilitator into something like Omegle—allowing random real-time conversations between Jewish and Palestinian participants—he chuckled at my enthusiasm but told me that he is interested in the possibility.

Perhaps that’s the most striking thing about Jaber’s approach: the combination of technical sophistication with his profound humility about what technology can and cannot accomplish. In building an AI that speaks with his voice, he’s created something that embodies both the promise and the limits of algorithmic mediation. It cannot solve the fundamental political questions that divide Israelis and Palestinians, but it might create spaces where those questions can be discussed without the conversation immediately dissolving into recrimination.