I've been curious about the results of the 2024 Presidential Election and the differences between the popular vote and the electoral college. I wondered why no one really challenged the final count—especially since Trump, back in 2020, claimed it was rigged. It appears that Electoral College members from Red States followed orders quite blindly, likely out of fear of the consequences. Meanwhile, the Blue States split their votes, favoring the Republican nominee. Trump secured the popular vote with about 49.8%, which always puzzled me because the margin wasn’t overwhelming—many people didn’t vote at all, possibly influenced by the presence of a woman candidate or political biases. I imagine some support from the Free Palestine Movement might have been influenced by their reaction to Biden and Trump. Ultimately, though, it seems we got the result we expected.
I came across a fascinating article in Forbes, written by Ron Schmelzer on September 25, 2025, discussing how AI models behaved similarly to voters during the 2024 election. Researchers from MIT and Stanford monitored 11 of the world’s top large language models (LLMs), including GPT-4, Claude, and Gemini. These models weren’t just answering questions—they engaged actively in conversations during four critical months of the 2024 presidential campaign. Over the course of the study, more than 12,000 structured questions were asked nearly every day from July to November 2024. It's a pioneering and comprehensive examination of how these AI models behave during real-world democratic events. The results? These models are more than neutral observers; they tend to react, sometimes show inconsistency, and can even be influenced by public narratives, even when they ideally shouldn’t be.
This indicates that chatbots can be easily influenced, not only by the questions people ask, but also by what they recall from previous conversations. However, the study couldn’t test chatbots with memory, mainly because it would have been too costly, as they would have needed thousands of accounts for the experiments. This highlights broader concerns about how models might drift over time due to their personalization. A chatbot with memory could become more biased simply by recalling past interactions, which could reinforce earlier impressions and amplify the effects of how prompts are written.
Reflecting on all this, it’s hard not to feel that these advanced technologies like LLMs and chatbots might have influenced the election’s outcome. I also think Trump and his party were aware of this, as he even mentioned once that he might have rigged the previous election. He's still closely connected to Elon Musk, who has developed AI systems such as Grok and xAI, and recently announced a deal with the government to allow federal agencies to use their chatbot for a small fee. Meanwhile, the Russians had their own chatbots and knew how to potentially sway Trump’s leadership for their own benefit. Social media helps connect the dots, leaving voters unaware of what’s really happening. Don’t forget about the voting machines, which can also cause issues with laws for redistricting and vote suppression. Currently, voters face numerous challenges and uncertainties. Who can we really trust? Big brother is watching us
No comments:
Post a Comment