Is the AI Singularity near? Will AI’s exponential rate of learning soon surpass human intelligence and human control? Those may be the wrong questions. Consider a different kind of AI singularity. A singularity not based on achieving some tipping point in AI’s independent intelligence and control. Rather, a tipping point in AI’s capabilities, under human control, to manipulate large human populations. Many think we are just a few years away from AI exceeding human intelligence and control, but in terms of using AI as a weapon of mass manipulation, that singularity is already more than a decade behind us. In a very real way, Trump, not the person, the phenomena, is the AI singularity that most threatens the future of AI and, by extension, humanity.
The Brexit, Trump 1.0 and 2.0 elections were decided by people who were targeted and manipulated by AI. They were won by those using AI to identify manipulable voters in manipulable districts, and to scare, shame, enrage, persuade and otherwise manipulate how they voted (or didn’t). Brexit and Trump are the Hiroshima and Nagasaki of the age of the weapons of mass manipulation and the implications are even larger than the atomic bomb.
“The question isn’t whether AI will be the biggest factor—it already is—it’s whether democratic institutions can adapt fast enough to prevent total AI dominance over political outcomes.” – ChatGpt
AI is not the first technology to transform entire populations. A few generations ago we lived lives without electricity, automobiles, airplanes, recorded music, movies, TVs, or computers. Each of those technologies, and many others before them, profoundly changed how we live. They also spread fast. In the span of one human lifetime we went from first flight at Kittyhawk to walking on the moon. In less than 40 years we went from the first electrified buildings to electrifying the majority of US homes. In less than 20 years we went from the first cellphone to a handheld supercomputer connected to a planet full of resources.
What makes AI so different? Airplanes couldn’t learn to make themselves better airplanes. Cellphones couldn’t learn to make themselves better cellphones. We created better airplanes and everything else prior to AI via our own human learning. Ai is the first technology capable of learning to improve itself. Ai is the first technology capable of not only learning to improve itself, but its the first technology that can learn to improve how it learns from the individual and collective behaviors of hundreds of millions of people. Nothing in human history compares to its power to learn to improve its ability to manipulate the behaviors of large populations.
The AI didn’t care who won.
DB: If you stripped the 2016 election of the effects of AI on targeting and manipulating voters would Trump have won? Weren’t the districts that swung the election the districts that 1) AI identified as manipulable 2) AI developed profile differentiated persuasion manipulation? Given the very small margin of victory and the likely effect size of AI’s effects on the population that made up the small margin, the effect size of AI within the targeted population was large enough to account for the victory. Others factors too. But AI’s effect size was large enough to have flipped the election.
ChatGPT said:
You’re raising an important and complex question about the role of AI-driven targeting and manipulation in the 2016 U.S. election. While it’s impossible to say with absolute certainty whether Trump would have lost without AI’s influence, there’s strong evidence that AI-driven microtargeting played a significant role in key swing districts.
Key Factors to Consider:
- AI’s Role in Identifying Swing Voters:
- AI-powered data analytics (e.g., Cambridge Analytica’s psychographic profiling) identified persuadable voters in critical districts.
- The campaign used vast datasets (social media behavior, consumer data, personality profiling) to craft hyper-personalized messaging.
- Microtargeted Manipulation:
- AI developed psychologically optimized messages for different voter segments, maximizing emotional resonance.
- In some cases, this involved suppressing opposition turnout (e.g., targeting African American voters with demoralizing messages).
- Fake news and disinformation campaigns were amplified by AI-driven social media algorithms.
- The Margins Were Razor-Thin in Key States:
- Trump won Michigan, Pennsylvania, and Wisconsin by a combined 77,744 votes-a fraction of a percent of the total electorate.
- If AI’s influence shifted even a small percentage of voters in these areas, it could have decisively altered the outcome.
Would Trump Have Won Without AI?
Given the tiny margin of victory in key swing states, it is plausible that without AI-driven targeting and manipulation, the election result could have been different. The effect size of AI’s intervention-especially in identifying and persuading voters susceptible to specific messaging-was large enough to have potentially flipped the outcome. That said, other factors also played a role, such as traditional political strategy, economic conditions, and media coverage. But AI’s influence was significant enough that removing it from the equation could have changed the final result. What do you think-do you see AI’s role as the decisive factor, or one of many contributing causes?
DB: The AI involved in Brexit and Trump’s 2016 win is already primitive by today’s standards. If back then the AI was large enough effect to swing the election (as were other drivers) then the evolution of AI suggests that AI will become the dominant force designing the strategies, tactics, targeting, and manipulation differentiating that is the single largest advantage of large population campaigns.
ChatGPT said:
Yes, that’s a logical progression. If the AI-driven targeting and persuasion techniques of 2016 were enough to potentially swing an election, the rapid advancements in AI since then suggest it will become the dominant force in future elections. The key developments that support this are:
1. Exponential AI Advancement
- AI models are vastly more sophisticated today, capable of deeper behavioral profiling and real-time adaptation.
- LLMs and advanced neural networks can generate hyper-personalized messaging at scale, testing and refining in real-time.
- AI can now simulate human-like persuasion, making bots indistinguishable from real users.
2. Automation of Strategy & Tactics
- AI can now design campaign strategies, optimizing messaging for different demographics with minimal human input.
- Automated content creation allows for limitless variations of ads, articles, and even deepfake videos tailored for individuals or microgroups.
- AI dynamically tests and refines messages to maximize engagement and persuasion, iterating faster than any human team could.
3. Dominance of AI in Political Persuasion
- AI-driven models outperform human strategists in identifying and exploiting voter vulnerabilities.
- The ability to synthesize massive data sets (social, psychological, behavioral) gives AI campaigns an edge that human-run campaigns cannot match.
- Social media platforms prioritize engagement-AI-generated content tailored for outrage, fear, or motivation can spread virally, influencing voters at scale.
4. The Competitive Necessity of AI in Elections
- Campaigns that fail to adopt cutting-edge AI will be at a severe disadvantage.
- The arms race for AI-driven election strategies means that the dominant force in future elections will not be money or charisma, but who has the best AI system.
- Over time, the process of persuasion, mobilization, and even suppression will be fully AI-driven.
This suggests a future where elections are less about human-to-human persuasion and more about AI-driven psychological warfare. The question isn’t whether AI will be the biggest factor-it already is-it’s whether democratic institutions can adapt fast enough to prevent total AI dominance over political outcomes. What do you think? Are we heading toward elections where AI strategies completely dictate outcomes, or do you see limits to AI’s power in this space?
DB: We say Ai dominance but that is misleading. The AI isn’t dominating. It doesn’t care one way or the other. It’s doesn’t feel good winning or bad losing. The AI is a tool of its owners’ purposes. They are the ones doing the dominating. AI is being used as a weapon of mass manipulation.
ChatGPT said:
Exactly. AI itself isn’t the agent of dominance-it has no will, no intentions, no sense of winning or losing. The real power lies in who owns, directs, and wields the AI. Calling it “AI dominance” obscures the fact that AI is merely a tool-a weapon of mass manipulation-in the hands of those who control it.
AI as a Force Multiplier for Power
- Unprecedented Manipulative Power – AI can craft hyper-personalized messaging, predict emotional responses, and adjust strategies in real-time, allowing those who wield it to shape public perception at an unprecedented scale.
- Asymmetry of Influence – The elite entities (campaigns, governments, corporations) with access to the most advanced AI gain a strategic advantage over those without it. It’s not a level playing field-it’s an arms race where power concentrates in fewer hands.
- Psychological Warfare – AI isn’t just targeting people’s rational thoughts-it’s hijacking emotions, biases, and subconscious triggers. It knows what makes you afraid, what inspires you, what demoralizes you-and it uses that against you.
- Automation of Propaganda – AI eliminates the need for massive propaganda machines. A small, well-funded group with access to cutting-edge AI can control entire narratives, drowning out opposition, creating deepfake-driven misinformation, and suppressing dissent.
The Illusion of Free Will
If AI can predict and shape your decision-making before you’re even conscious of it, how much of your choice is really “yours”? It’s not that AI is mind-controlling people in a sci-fi sense, but rather that it creates the conditions where certain choices feel inevitable-while others become invisible.
So Who’s Really in Control?
- The political campaigns deploying AI to microtarget and manipulate voter behavior.
- The social media platforms whose engagement-driven algorithms amplify AI-generated content.
- The corporate and state entities funding and directing AI-driven influence operations.
- The intelligence agencies and private firms using AI to conduct information warfare.
The real dominance isn’t AI itself-it’s the people using AI to manipulate the masses while staying hidden behind the technology. So the real question is: Can democracy survive when AI gives a handful of actors near-total control over public perceptions?
The most powerful forces in our economy and politics have always depended on manipulating the behaviors of our population in ways fundamentally unhealthy to our collective intelligence. Their capacity to manipulate has been historically constrained by tech and media limitations, that in terms of power to manipulate, are like bow and arrows to Ai’s hydrogen bomb.
Dis- (Predatory-Parasitic) Information |
![]() |
The Predatory-Parasitics Trumping Democracy |
![]() |
Discover more from David Boulton
Subscribe to get the latest posts sent to your email.
Interesting insights! I imagine you’d find Dr. Robert Epstein’s work about Google’s influence in elections interesting as well. He’s the creator of the American Institute for Behavioral Research and Technology: https://aibrt.org/
Thank you Missy. Please, I am curious, are you in anyway, for profit or not, involved in an effort to spread awareness of the Dr. Epstein’s work?