Quarter of Israelis Consider AI for Voting Decisions, Survey Finds

A comprehensive survey by the Israeli Internet Association reveals that approximately one-quarter of Israelis are considering using artificial intelligence tools to help them decide which party to vote for in the upcoming 26th Knesset elections. The survey, conducted by Geocartography among 1,005 Israelis, also found that a majority of the public has encountered false information related to the elections, and most are concerned about escalating online political discourse.

According to the findings, 69% of the public expressed significant concern that the political discourse surrounding the election campaign will lead to increased online violence. This fear is heightened by widespread exposure to disinformation, with about 71% reporting frequent or occasional encounters with false information, and only 15% stating they have not encountered it at all. The issue of online impersonation is also prevalent, with 56.8% reporting frequent or occasional encounters with fake accounts or impersonators.

The survey further indicates a significant presence of malicious AI use, with 48.2% reporting frequent or occasional encounters with AI used to harm or defame others. The Israeli Internet Association emphasizes that impersonation, disinformation, and harmful AI use are not future threats but current realities of the online space. They warn that existing digital risks, including violence, incitement, and disinformation, could intensify during the election period.

Notably, the disinformation encountered is not perceived as belonging to a single political camp. The survey showed similar exposure rates across the political spectrum: 73.3% of right-wing respondents, 69.5% of centrist respondents, and 68.9% of left-wing respondents reported encountering disinformation.

In response to these findings, the Israeli Internet Association has published a guide for voters on how to deal with false information online. The guide advises checking the source of information, preferring credible sources, cross-referencing information, and verifying publication dates and context. While AI tools can be helpful for learning and comparing stances, the association stresses that AI responses are not objective recommendations and can be flawed, outdated, or even fabricated. They recommend using AI for research and comparison rather than asking it directly for voting advice, suggesting prompts that focus on comparing party stances, providing sources, and presenting arguments without endorsement.