The Virtue Silo: Why Liberals Don't Want to Understand AI

I want to be clear on my position: technologists are inherently political actors in the modern world. From Oppenheimer to Fei-Fei Li, who recently warned about AI misinformation and the risk of harmful overregulation [1], builders of technology have shaped some of the most politically consequential acts of our time. Social media is a primary example, with both primary impacts (those who control data and communication shape perception, erode democratic norms, and enable authoritarianism) and secondary ripple effects. The Cambridge Analytica whistleblower revealed in 2018 how Facebook data was extracted for campaigns against specific voting blocks in the US election [2].

So what does this have to do with liberals? Despite the obvious political power of technical systems, liberal spaces systematically decouple the inherent design architecture from the emergent consequences humanity faces and prioritize the resulting social media phenomenons: 'intellectual virtue signaling' [3] and moral virtue signaling. This toxic behavior has created a meta Dunning-Krueger effect across the entire western progressive ecosystem which is seeing a super charge with artificial intelligence. Most liberals became a main character with Instagram and took a moral position on every relevant topic with but also with a curious pseudo-expertise. This inflation of our abilities and credentials was easily fed by the ubiquitous passing of buzzwords and 'pop knowledge' (shallow viral replacement to grounded expertise) in social media. Propagators were rewarded with re-posts, likes, story shares, and comments and a tribalism took over. The siloed echo chamber had its way with reality. Nuance lost its footing, and radicalization gained traction.

Social media trained liberals to perform. AI now demands that they actually understand without reward. The two are incompatible. Artificial intelligence is a very specific tool, but one with many different designs and deployments. Each person should do cursory education with what it is. The sleight of hand of ungrounded 'impressions' doesn't hold up with AI. You must know what it is and that means doing something that... is not cute. Reading 'Attention is All You Need [4]' and thinking of how to implement human values into algorithmic structures is boring and unglamorous work. It really harkens to Catherine Liu's Trauma, Virtue, and Liberal Elites [5] podcast appearance, where she discusses how liberal elites prefer to criticize from a safe place of privilege but not do the meaningful work. This is because the idea of something became replacement to the thing itself in the psychosis of the convergence of social media and reality spaces. The performance of virtue is much more alluring than rolling up your sleeves and doing unrecognized work that defines said virtue.

So, in this virtue silo, liberals don't want to understand AI. First, it means they have to admit there are some really complicated things they may not understand and have to sit with and process. This is uncomfortable and the tribalist echo chamber has not allowed for the nuance of this type of conversation. People speak at each other, and support themselves with self proclaimed authority and expertise. Second, it's boring and not self-centered. It's like going to Erewhon to get a smoothie or a National Park to get that IG pic, you want to go where you will be seen because perception > reality. Lastly, liberals have lost the art of disagreement and debate in a way that polices conversations, thoughts, and actions and restricts any dissent, novelty, or intersection.

Since taking on the challenges of how AI can be used requires doing all of these things, I notice liberals are choosing to just be against or highly critical of AI all together, as if artificial intelligence is a monolith and there is no complicated technical conversation further into the topic. They don't want to discuss technical architectural changes or shifts and roll up their sleeves and make solutions. I see criticism after criticism, symposium and echo chamber after echo chamber, none integrating the necessary technical voices, complexity, and nuance that is required for meaningful solutions. It's not the fault of AI we are in late stage capitalism everyone, calm down.

Take the Civil Society Summit on AI, held in Montreal in May 2026. Its panels covered algorithmic harm, digital sovereignty, autonomous weapons, and environmental justice. Its speaker list included exactly one person with a technical AI background — the rest were lawyers, policy experts, activists, and academics from the humanities. A symposium on AI with almost no one who builds AI. This is really common practice in these conversations.

But the summit's missing technical voices are not its only blind spot. The environmental concern, while not irrelevant, is overstated. I also see this everywhere now. There are much bigger contributors to global warming than AI data centers, and the pace of demand is likely a bubble that will correct itself. AI's energy efficiency is improving by roughly an order of magnitude each year — and AI itself has the capacity to increase energy efficiency across other sectors, as documented in the IEA 2026 report [6]. The point is these are not black and white issues, they are complex projects that require the technical discipline to be present in the conversation. There are analytics and numbers that can ground claims...

I am currently working on building new tools and not engaging with social media (for the most part). I love what I am doing, but I fear what Fei-Fei Li flagged -- liberals do not want to understand AI and that ignorance can have an incredibly harmful consequence for society overall. I hope we can burn the performance and start actually building with humility, because 'Big Tech' isn't going to stop doing it.