Can AI Decide the Future of Our Elections?

Gain a better understanding of Artificial Intelligence, its diverse applications, and its impacts on education, healthcare, law, etc. For this post, I want to focus on something very relevant currently—the Elections, and how AI will play a role in that. I also want to give a comprehensive review of the limits and possibilities of AI in general since this is the last post.

Strategy:

I have continued to follow two magazines focusing on new developments in AI:

·       AI Magazine by the Association for the Advancement of Artificial Intelligence (https://aaai.org/ai-magazine/)

·       Communications of the ACM—Artificial Intelligence and Machine Learning (https://cacm.acm.org/category/artificial-intelligence-machine-learning/)

I have also followed these podcasts:

·       Mind Matters (https://mindmatters.ai/podcast/)

Current Reaction:

In no time, I have reached the end of the semester and the last CLP post. I have learned a lot so far about AI, and the different fields it is impacting. Even after the class ends, I will continue to learn and follow these magazines, as I find it a very fulfilling pastime. Recently, I have been fascinated by both AI’s practical uses and philosophical concerns. AI-generated art and video are becoming commonplace across different social media platforms and making people, myself included, question their reality. When I see any infographic or sophisticated image, I wonder whether this is factual or the result of AI? When I see the election predictions and surveys, I have the suspicion that these are manipulated by AI.

Sources:

·       Communications of the ACM article, “Keeping AI Out of Elections” 

·       AI Magazine by the Association for the Advancement of Artificial Intelligence 2024 Conference Reports

·       Mind Matters Episode 293: THE LIMITS AND POSSIBILITIES OF ARTIFICIAL INTELLIGENCE

Synopsis:

The integration of AI in elections is a double-edged sword, presenting both opportunities for enhancing electoral processes and challenges in terms of misinformation and manipulation. The "AI for Credible Elections" workshop at AAAI-2024 gave me a good idea about the constructive use of AI to address electoral challenges. Experts from fields such as cybersecurity, journalism, and political science discussed how AI can provide accurate, personalized voting information, improve election monitoring, and combat misinformation. Professor Matthew Saltzman highlighted the role of platforms like Vote411 in recognizing reliable voting information in the U.S., while discussions also covered the use of crowdsourcing platforms like Ushahidi for election monitoring in Africa. However, there were concerns about information disorder, with presentations on technology aiding journalists and tackling fake news. The panels at the workshop debated the pros and cons of AI in elections, emphasizing the need for authoritative data sources, standardized AI practices, and increased funding for AI innovations in elections.

On the other hand, the article, “Keeping AI out of Elections”, showcased the potential dangers of AI in electoral contexts. Jennifer Golbeck, a computer science professor, emphasized how social media platforms' inconsistent enforcement of their policies during the 2020 U.S. elections led to widespread misinformation. The session pointed out that generative AI tools can amplify both positive and negative content, needing strict rules to manage AI-generated misinformation. Legal experts like Jessica Selinkoff from the Federal Election Commission discussed the limitations of current regulations in addressing AI's role in political campaigns and the urgent need for legislative updates. Vivek Krishnamurthy highlighted global examples of AI misuse in elections, such as AI-generated videos in Pakistan's elections, and stressed the complexity of implementing effective regulations quickly. Despite these challenges, some tech companies have started initiatives to combat AI interference, and there's a strong call for supporting local journalism to produce accurate content. Ultimately, the article gave the shuddering realization that these factors have to be controlled before it is too late in the elections.

Analysis:

I recently listened to an episode from the Mind Matters podcast, which very much summarized my doubts and hopes regarding AI. Through the perspectives of George Montanez, Robert J. Marks, and William Dembski, I learned that generative AI models, despite their impressive capabilities, are fundamentally correlation-based systems that rely heavily on human rationality, creativity, and biases. Montanez emphasized that these systems can absorb and interpolate data to create outputs, but they quickly degenerate when trained on their own outputs, a phenomenon known as "model collapse." This highlighted the significant differences between human and AI rationality, making me skeptical about AI ever achieving human-like consciousness.

Robert J. Marks opened up a discussion on the computational limits of AI, contrasting the vast amounts of data and computational power required to train AI models with the minimal input humans need to learn and perform tasks. This realization made me appreciate human cognitive abilities even more. Meanwhile, William Dembski's skepticism about AI surpassing human capabilities resonated with me, particularly his argument that AI's reliance on large data sets contrasts sharply with humans' ability to derive significant understanding from minimal input. Dembski's concerns about the societal implications of over-reliance on AI made me reflect on the potential dangers of undervaluing human capabilities and reducing ourselves to machine-like thinking. Overall, the podcast made me excited about AI's potential while also cautioning me about its limitations and the ethical implications of its integration into society.

Insight:

I learned that while AI models are impressive and have potential applications in areas like predicting election outcomes, their reliance on vast data sets and human biases raises concerns about their limitations. This realization makes me question our dependence on AI and after reading my past CLP posts I can see that AI is already seeping into every possible field. Ultimately, with my pessimistic feelings about humanity being taken over by AI, there is new hope after all this research and the fact that we know AI’s limits.