The Accelerant Effects of AI

A survey on social justice issues in a state senate district found that concerns about data center construction tied with concerns about aging gracefully, reflecting broader societal anxieties about AI's impact. The author argues that AI acts as an accelerant, amplifying existing trends and power dynamics, making creative work more valuable while threatening vulnerable workers.

I’ve had now hundreds of conversations about AI. Surprisingly, rather than being a simple productivity tool, these conversations meander into debates and speculation about social change, and reveal our greatest fears and deepest desires. Simultaneous to my vocational conversations, I facilitated running a survey on social justice issues in my state senate district. To my surprise, concerns about data center construction were tied with concerns about aging gracefully. A big question on a lot of people’s minds seems to be “what do we actually get for all of this”, and it’s a fair question. Equally interesting to me is the diversity of reactions, which I’d like to explore here. In an earlier era of AI tools, I was hooked by fascination with the ability of the tools to map concepts to mathematical vectors. The canonical example of this was “king – man + woman = queen”. With some effort, AI’s capability allows you to search text for concepts, without necessarily using the correct terms. If you read my post history way back, I explored ways of using this, and built a search engine for lectures which allowed me to find great thinkers who who would otherwise be unknown to me. A model built on concepts also gives tools the power to summarize: if a thought is constructed in a meandering fashion, the same idea can be represented concisely. This goes other ways too: a concise thought could be obfuscated, as well translated from the lingo of one subculture to another. In software development, we are still struggling with how to express “it” of craftsmanship. A piece of code that satisfies a feature can be represented many ways: from terse and lacking error checks, to meandering and overly verbose. Somewhere in between there is a sweet spot, with code that is robust to reality, but focused and resilient to future needs of it’s owner. The capacity to summarize and de-summarize also makes AI into a powerful accelerant. For the types of work it touches this affects societal phenomena well beyond just “coding”, and I would propose that it explains the variety and intensity of all of our reactions. For someone with a lot of agency to be creative, a tool that accelerates this is very compelling. For a self-reflective leader, the tools can find blind spots and skill up quickly. In contrast, for someone who is vulnerable to abusive systems of power, or is in danger of losing the most fun part of their work, the tools are deeply unsettling. There is a business theory about “complementary” goods that fits well here. There are many things which “go together”: pasta and sauce, cars and gasoline, beauty and wonder. If something becomes easier to obtain, the other item in the pair becomes more precious. Simultaneously, the “cheaper” item in the pair gets more use. This is visible in both our day to day work, and at a large scale. In a workplace “technical debt” work things that are just barriers to the direct work , become simultaneously cheaper to staff and more valuable to address, because they prevent both humans and AIs from working. Skills like choosing the the right problems to solve, and steering a team towards them become more valuable, even though they were always important. Turning the wheel of a slow-moving 1950s tractor has a different impact than shifting the path of a commercial airliner. I suspect we will see areas where this accelerates other phenomena as well: technology allows people who are lonely or anxious to become moreso my mental health improved when I dropped instagram . I have also seen that people who are finding ways to cope with learning disabilities or communication challenges are able to do so much easily. Tool vendors advertise that they can replace human labor with machines. While I find the claim dubious, I think the message remains clear – powerful people and institutions see in AI a tool that allow for the furtherance of their self-interest, and futher concentration of wealth and power. Earlier in the year, I watched a presentation by the PA department of health and human services. There are a lot of national changes reductions in funding, both in the amounts and the accessibility of funds. One thing here caught my eye- funding cuts for the “most difficult to house” people sex offenders, persons with drug additions, severe mental health issues is expected to lead to 5500 additional homeless people in PA. This coincides with the AI message in a key way: more visible social problems present a reminder to the rest of us of the worst of what can happens if we don’t find a way to act in the interests of the power players of the world. Ultimately all of this makes things our humanity more important and precious. There are many things LLMs cannot do. Those things were significant in a world without AI, but they become more special because of their rarity, and we should change our efforts to pursue them: the ability to form relationships, intuit and influence power, and to create spaces filled with beauty, meaning and a sense of belonging. In closing, I am pleased to see that some Pennsylvania legislators responding to the moment. My state representative Rep. Melissa Cerrato voted in favor of a digital ad tax in PA HB 1678 https://www.palegis.us/house/roll-calls/summary?sessYr=2025&sessInd=0&rcNum=1143 , and my state Senator Sen. Maria Collett signed onto a memo https://www.palegis.us/senate/co-sponsorship/memo?memoID=48849 calling for an end to PA’s regressive tax system. While neither of these are passed into law, these are important first steps to “test the waters” and see what is possible to support their constituents.