Perhaps the Flaws aren’t with AI, They’re With Us

By Julian | Last Updated: 14 October 2025

Now before you think this is paid for by a big tech and that I’m trying to let them off the hook. I’m not. AI is built and used irresponsibly from its carbon footprint to its complicity in user deaths. However, what I will argue is that the areas where we are seeing problems with people interacting with AI is where we already face genuine societal problems.

Jobs

Even before LLMs and AI generally stoked fears of replacement, a culture of routine firing did. Since the 1980s US firms have regularly conducted mass layoffs to make short term gains for investors. Worker security is a thing of the past. The simple fear is stoked by a business reality: if a robot can do the job cheaper why hire a human? It hasn’t been that long since slavery was widespread: owners will always want free labor. In Europe there is better job security but still the threat that entire roles can be made redundant is intact. Europe is ever at the precipice of cutting down its social safety net and worker protections for the appearance of being competitive. Right now it is unclear if AI will concretely replace jobs, however what’s important is that companies think it can. There are fewer and fewer entry level positions available, the workplace is being restructured.

PBS has lost the P

The media environment particularly online, is said to be under threat from AI generated websites, deepfakes, audio impersonations, and even swarms of Agentic AI. However, what the debate on these threats often misses is that the reason people might fall for these false reports: The traditional media environment has failed to be reliable. Financial support for independent journalism is low. This paves the way for the end of reporting in some areas or worse, buyout of news outlets by rich individuals with corporate interests (Bezos, Bloomberg, Murdoch etc.). A rich diverse media environment limits the holdings of individuals in disseminating solely their version of truth. A strong education system gives people the tools to identify where they’re being manipulated, both by traditional news outlets and AI agents. Speaking of education….

Civics

The worst nightmares of many of the US Founding Fathers have come true. The rabble are running the show. Everyone can vote, everyone has a voice. If they’re taught how to use it. Civics education in its current state is abominable. In the US over 70% of Americans fail a basic civic literacy quiz. In the EU people tend to trust institutions (51% trust the EU) however awareness of the what the EU is doing and how it works is patchy and varies greatly from country to country. Democracy demands more than showing up at the ballot box every couple of years and picking your least unpleasant flavor of ice cream. It’s about participation, the palpable feeling that you are a part of something and that your voice can be heard. Additionally, it needs to be about more than voting for the betterment of your wallet. Democracy requires a forum of genuine citizen participation and debate. The same technological tools that divide us in rage filled comments can be used to find consensus.

Even basic literacy rates leave room for improvement with 48 million Americans unable to read above a 3rd grade reading level. While in Europe one in five are considered low performers. We need education that works for everyone. A democracy is only as good as it’s citizens understanding of how it works.

Looking for Friends

We live in lonely times In the US a CDC survey found 1 in 3 U.S. adults report feeling lonely, and that about 1 in 4 report not having social or emotional support. In the EU survey data found 13% of respondents report feeling lonely most or all of the time over the past month and 35% report being lonely at least some of the time. And there are people for whom LLMs are a safe place to explore socializing without fear of judgement or ostracism. However LLMs are by nature sycophantic. If you want a full list of why AI is an inadequate substitute for human contact check out Onno’s piece. The consequences of LLM inadequacy for supporting vulnerable people unfortunately has been played out in a number of tragic suicides. We need to find ways to generate community and make people aware of the dangers of developing attachments to LLMs.

What’s to be done?

A one size fits all solution as always will not work. The EU’s AI Act needs to be supplemented with initiatives that address the underlying harms that AI causes. We need to invest in people as much as we invest in AI. People need safety nets when in danger of occupational replacement, resources for re-education if their sector is impact. But on a deeper level we need people to have the time to think critically, to be supported to do in-depth research on their democratic systems and how they can move their interests forward and most importantly to be able to be there for each other. Each of these is a mammoth undertaking, but humanity is always worth investing in.

Get the latest updates