The Outsider Perspective Issue 396 | The Beltway Outsiders

If you’d like to read this issue on my website, click here! If you’d like to sign-up, and receive this in your inbox each week, click here! Read past issues here.

Good Friday Morning! Except for George Frandsen, who is in the Guinness World Records for the… and I quote… “largest collection of fossilized feces” in the world. He’s now opening a museum of all his finds in Williams, Arizona. I appreciate the name: the “Poozeum” is open to all and has received an “enthusiastic response,” according to the New York Post.

I have an announcement this week: I have started a YouTube channel called The Horse Race, where I discuss the latest polls and where the Presidential race is heading as we near November.

It’d be helpful if you subscribed to the channel and liked the videos as they come up. It’ll help push the show to others. There were some audio issues at the end that I’m aiming to iron out for future shows.

This week, I’m going to discuss how artificial intelligence is reshaping politics. Would you believe me if I told you AI could replace humans in polls? Maybe not yet, but you will by the end of this newsletter—links to follow.

Quick Hits:

CNN ran a hit piece on Tucker Carlson and Ticketmaster this week. And you know the piece was terrible when I’m sitting here on the side of both Ticketmaster and Tucker Carlson. Tucker is headed on a 15-city arena tour this year, and Ticketmaster is selling tickets (as they would anywhere), and CNN wants Ticketmaster to shut the events down. The “journalist,” Oliver Darcy, spent a week pestering Ticketmaster and all the sponsors of the venues in those places, asking why they were 1) selling tickets for Tucker and 2) letting Tucker run events there. No one responded. Again, this is allegedly a journalist, not an op-ed writer, and he puts this segment in his piece: “Companies with basic ethics should, of course, reject doing business with dishonest figures who profit by dumping toxic waste into the country’s information environment.” Here’s how you know Darcy is lying about this point: ever since October 7th, we’ve watched a wave of leftist antisemitism crash over the West. These groups continue to get funded by liberal groups, and some of them even speak on CNN as contributors. In 2019, CNN invited self-avowed white nationalist Richard Spencer onto their network for the sole purpose of trashing Trump (that was not Spencer’s first time on CNN). Darcy isn’t doing this because of rhetoric; he’s doing it to try and impact the election. These kinds of excuses are being drummed up to explain a possible Biden loss in November. It’s no wonder CNN is collapsing from poor ratings.
The Washington Free Beacon dropped a bombshell report on Columbia University. My jaw was on the floor. During a panel where administrators listened to alumni and others on problems relating to antisemitism at Columbia, several leaders of Columbia were caught trashing the testimony in text chains to each other. The texts were found out because a person in the audience was sitting behind these administrators, taking pictures of the texts they were sending. The Free Beacon summed it up, “Throughout the panel, which unfolded over nearly two hours, Chang-Kim was on her phone texting her colleagues about the proceedings—and they were replying to her in turn. As the panelists offered frank appraisals of the climate Jewish students have faced, Columbia’s top officials responded with mockery and vitriol, dismissing claims of antisemitism and suggesting, in Patashnick’s words, that Jewish figures on campus were exploiting the moment for “fundraising potential.” The report has pictures of the texts being sent that have been independently verified. No one involved had a statement.
College athletics is moving into a new era. The BIG 12, now without Texas and Oklahoma, is working on securing “a first-of-its kind private equity investment to ensure the league’s long-term financial and competitive security.“ CBS News reports, “On the table is a possible cash infusion of $800 million to $1 billion from Luxembourg-based CVC Capital Partners in exchange for a 15% to 20% stake in the league.“ This deal will not be the last. Over the previous year, Josh Pate at Late Kick has floated the possibility of Middle East/Saudi Arabia money entering the equation for some of these leagues, too. The Saudis, in particular, are working hard to pivot away from oil-dependent investments by 2030. The monetary arms race in college athletics is a prime candidate to draw in these investments. A European firm popping the bubble likely ensures a flood of foreign investment into college sports.

Where you can find me this week

Please subscribe, rate, and review my podcast on iTunes, Spotify, or Google Play — the reviews help listeners, and readers like you find me in the algorithms. Make sure to sign up for the Conservative Institute’s daily newsletter.

The Far-Left Blasts Israel’s Heroic Rescue Operation – Conservative Institute

Liberals Are Reinventing Law Enforcement On The Fly – Conservative Institute

Should Liberal Supreme Court Justices Resign? Liberals Think So… – Conservative Institute

The New AI Concept: Replacing Humans In Political Polls

If I’ve beat one drum over the last year, it’s been this: artificial intelligence is reshaping everything. There is not one untouched facet of our lives that AI is rushing into at full speed. Is this good? Bad? I can’t tell because the pivot is occurring at light speed.

This week, I saw a use case of AI that I never anticipated, but suspect will multiply: using AI to replace public opinion polls.

The idea comes from the Harvard Kennedy School – Ash Center for Democratic Governance and Innovation. They published a piece titled: “Using AI for Political Polling: Will AI-assisted polls soon replace more traditional techniques?“

Before you laugh or roll your eyes, please realize that they’re publishing this not as a theory but as a research-tested concept.

To explain how they’re doing this, I need to define a concept I haven’t discussed here before: AI Agents.

What Are AI Agents?

Artificial Intelligence (AI) agents are like smart assistants created by humans to help with specific tasks. They are trained to become a specific “agent” to act as a virtual version of a real person. Think of them as virtual robots that can learn and make decisions. You can create an AI Agent that works as your personal assistant or acts like an editor for a top publisher. The AI will view requests through the lens that you use to request them.

If you’re designing an AI agent to act like you, its decision-making is supposed to mirror yours. If you create one to be a writing editor, it’s supposed to analyze like an editor.

What Do AI Agents Do?

AI agents perform a variety of tasks, including:

Learning: They can learn from data and experiences to improve their performance over time.
Decision Making: They can make decisions based on the information they have.
Interacting: They can communicate with humans or other systems to get or give information.

How Are AI Agents Trained?

Data Collection: First, AI agents need a lot of data. This could be anything from pictures and videos to text and numbers. An AI Agent needs data that is particular to the job it is being asked to do.
Training: The AI is fed this data and uses it to learn. This process is like studying for a test. The AI looks for patterns and relationships in the data.
Testing: After learning, the AI is tested with new data to see how well it performs. This helps to check if it has learned correctly.
Improvement: If the AI makes mistakes, it gets feedback and adjusts its learning process to improve over time.

Examples of AI Agent Uses

Virtual Assistants: Programs like Siri, Alexa, and Google Assistant that help you with tasks like setting reminders or answering questions.
Customer Service: Chatbots on websites that help answer your questions or resolve issues.
Self-Driving Cars: Vehicles that can drive themselves by understanding the road and making safe driving decisions.
Recommendation Systems: Suggestions you get on Netflix or Amazon based on what you like or have bought before.
Healthcare: AI agents can help doctors diagnose diseases or recommend treatments based on patient data.

Returning to the polling question, the goal of researchers is to create AI Agents to mimic the answers of a real person who could answer a phone poll. Why would they want to do this? Two reasons.

First, it’s becoming impossible to get real people to answer phone calls for polls or return mail surveys. Harvard notes, “Pew Research reported that 36% of the people they called in 1997 would talk to them, but only 6% by 2018. Pollsters worldwide have faced similar challenges.”

Additionally, online panel polls haven’t proven to be all that useful. And people are not responding truthfully to polls (think shy Trump voters or anyone with extremist views). Sometimes, people answer not what they believe but what they think will anger the political party. Polling is struggling, and it’s getting more expensive. The few people that answer means you have to make more calls to get to a viable sample.

Why would Harvard or anyone else be interested in AI polls? Because AI will always answer questions. And compared to making thousands of phone calls, it’s very cheap. If you’re trying to get into the ballpark of a polling question (“am on on the majority or minority view of this public policy?”), It will be cheaper to use AI to find an answer.

With these things in mind, I’m going to quote the Harvard piece at length here because they explain the pros and cons of their “AI polling”:

When we ran our own experiments in this kind of AI use case with the earliest versions of the model behind ChatGPT (GPT-3.5), we found that it did a fairly good job at replicating human survey responses. The ChatGPT agents tended to match the responses of their human counterparts fairly well across a variety of survey questions, such as support for abortion and approval of the US Supreme Court. The AI polling results had average responses, and distributions across demographic properties such as age and gender, similar to real human survey panels.

Our major systemic failure happened on a question about US intervention in the Ukraine war. In our experiments, the AI agents conditioned to be liberal were predominantly opposed to US intervention in Ukraine and likened it to the Iraq war. Conservative AI agents gave hawkish responses supportive of US intervention. This is pretty much what most political experts would have expected of the political equilibrium in US foreign policy at the start of the decade but was exactlywrong in the politics of today.

This mistake has everything to do with timing. The humans were asked the question after Russia’s full-scale invasion in 2022, whereas the AI model was trained using data that only covered events through September 2021. The AI got it wrong because it didn’t know how the politics had changed. The model lacked sufficient context on crucially relevant recent events.

We believe AI agents can overcome these shortcomings. While AI models are dependent on the data they are trainedwith, and all the limitations inherent in that, what makes AI agents special is that they can automatically source and incorporate new data at the time they are asked a question. AI models can update the context in which they generate opinions by learning from the same sources that humans do. Each AI agent in a simulated panel can be exposed to the same social and media news sources as humans from that same demographic before they respond to a survey question.This works because AI agents can follow multi-step processes, such as reading a question, querying a defined database of information (such as Google, or the New York Times, or Fox News, or Reddit), and then answering a question.

In this way, AI polling tools can simulate exposing their synthetic survey panel to whatever news is most relevant to a topic and likely to emerge in each AI agent’s own echo chamber. And they can query for other relevant contextual information, such as demographic trends and historical data. Like human pollsters, they can try to refine their expectations on the basis of factors like how expensive homes are in a respondent’s neighborhood, or how many people in that district turned out to vote last cycle.

The pros are easy to see. If you’re trying to get a quick and easy opinion poll, this might get you into the ballpark. The drawbacks seem apparent, too: AI cannot anticipate politics shifting. It depends heavily on a static environment of beliefs.

AI assuming the left would be against supporting the Ukraine war is a telling point. It shows two things: 1) The negative polarization of politics (the anti-war left can’t talk about Ukraine), and 2) How politics is less about hard policy and more about partisanship.

I’m not anticipating AI polls being rolled out during this cycle. These things are still being tested at this stage. However, a pilot run in the 2026 midterms, followed by public AI polls in 2028, is a distinct possibility. I’d look for an “AI Polls vs. Real Polls” set up to become a thing. Pollsters might use AI to “check their work,” so to speak.

The big takeaway is that politicians might start looking to AI to test possible policy proposals and election slogans. In the copywriting world, you often do what’s called “A/B” testing, where you test various slogans against each other to see a response. An AI agent could let you test that much cheaper than in the past.

The Harvard research was done with ChatGPT 3.5, an older model. I’ve found GPT4 and the newest GPT4o to be far better and more accurate. It wouldn’t shock me that Harvard got better results with the newest models. Everyone expects ChatGPT5 to drop sometime in the next 6-12 months.

It’s a Brave New World. Think about that for a moment: there’s a growing belief that we can replace human input in polling to find out what the general populace wants.

If we can replace humans in polls, what won’t be replaced by AI?

I’ll leave you with this quote from the story: “Today, humans fill out the surveys and computers fill in the gaps. In the future, it will be the opposite.”

Links of the week

Sheryl Sandberg: ‘I Was Wrong About Antisemitism’: The former Facebook executive talks about the silence of the feminists in the wake of October 7. – Bari Weiss, The Free Press

Heroism and the Biden Brainless Trust – John Podhoretz, Commentary

I Went to Cover a Protest. I Was Surrounded by a Mob: But in the context of other incidents over the past 24 hours, what happened to me was mild. – Olivia Reingold, The Free Press

Progressives Realize They Have a Jew-Hate Problem: As the woke of New York issue condemnations of antisemitism, it’s tempting to dismiss it as too little, too late. But that would be a mistake. – Suzy Weiss, The Free Press

On Derek Chauvin, George Floyd, and Reasonable Doubt: Coleman Hughes wrote a column in these pages that set off a flurry of criticism. Here, he sets the record straight. – Coleman Hughes, The Free Press

How Public Schools Became Ideological Boot Camps: In nearly every public school in the country, children are given curriculum materials that have no official oversight or approval. – Robert Pondiscio, The Free Press

A Doctor Told the Truth. The Feds Showed Up at His Door: A young surgeon revealed how his hospital was secretly transitioning gender-distressed children. He is now facing federal charges. – Emily Yoffe, The Free Press

Why 270 is the most dangerous number: Biden needs a more robust electoral map. – Nate Silver

Joe Biden Is a Lousy Politician – Matthew Continetti, Commentary

Patagonia funneled thousands to Palestinian terrorism-linked group, documents show – Washington Examiner

Texas megachurch pastor Tony Evans resigns after nearly 50 years over mysterious ‘sin’ – NYPost