I became consciously aware of the idea that the internet should be treated like a utility during the net neutrality battles of the 2010s. That was the period when access, throttling, and gatekeeping became visible to me, not just as technical issues, but as questions of power. It was also impossible for me not to absorb that framing living alongside a network engineer for many years.
At the time, it felt like a new argument. In hindsight, it wasn’t.
What was new wasn’t the idea—it was the realization that something foundational was slipping.
The earliest versions of the internet were built in public institutions, funded by governments, shared across universities, and shaped by people who assumed access rather than debated it. The network that became the internet was closer in spirit to a library or a public road than a marketplace. Commercial ownership came later. Arguments about who gets access came later still.
Net neutrality didn’t introduce the belief that the internet should be a utility. It surfaced the fact that many people had always understood it that way and were now being asked to defend that assumption out loud.
Seen this way, the United States 2010 Net Neutrality debates weren’t about innovation. They were about memory. About whether we remembered what the internet was going to be about openness instead of extraction.
I’m starting to feel a similar moment now with advanced language models, the point where something that feels like it should be infrastructural is being treated as optional, premium, earned, and a profit center.
Not because they’re impressive (they are), or because they’re new (they won’t be for long), but because they are quietly becoming infrastructure for thinking. The kind that reshapes who gets to participate, who gets to keep up, and who gets left behind.
The question isn’t whether large language models should exist. They already do.
The real question is whether we treat them as luxury products, or as shared tools for public life.
Thinking tools, not magic machines
When people talk about AI, the language often drifts toward spectacle: intelligence, replacement, automation, disruption. That framing does real harm. It makes the technology feel distant, elite, and vaguely hostile, especially to people who already feel excluded from “tech.”
But most everyday uses of language models are not about replacing thought. They’re about supporting it.
They help people:
- organize ideas
- lower the barrier to writing
- ask better questions
- summarize dense or inaccessible text
- plan, reflect, and iterate
That makes them less like oracles and more like assistive literacy tools. Which means access isn’t a bonus feature, it’s the ethical center.
Access doesn’t live behind logins
Equitable access isn’t solved by subscriptions or free trials. It doesn’t even live primarily online.
Historically, when society has faced tools that dramatically expand literacy or participation, we didn’t hand them out individually and hope for the best. We built places. We built schools. We built libraries. We built public institutions where tools were shared, supported, and contextualized.
Not everyone owned a printing press. But rooms were built where knowledge was held in common.
If advanced language tools are going to shape education, healthcare, civic participation, and creative work, then access needs to be anchored in public, physical, human spaces—libraries, community centers, disability resource hubs, schools—not just private devices and paid accounts.
Disability makes the stakes clearer
For many disabled people (myself included), language models already function as assistive technology.
They reduce cognitive load.
They help translate bureaucratic language.
They support executive function, planning, and communication.
They offer a way to participate when energy, mobility, or stamina are limited.
When access to these tools is framed as a luxury, disabled people are quietly told that accommodation is optional. That’s not a hypothetical future concern, it’s already happening.
If we take accessibility seriously, then we have to talk about language models the way we talk about screen readers or captioning: not as indulgences, but as supports.
Use literacy matters as much as access
Giving people access without guidance creates a new kind of inequality.
A real divide is between those who have AI and those who don’t, but it grows even between for those who know how to question it, check it, and collaborate with it, and those who are may trust or fear it blindly.
What people need isn’t technical mastery. It’s:
- how to ask good questions
- how to verify and revise
- how to recognize bias and gaps
- how to protect their privacy
- how to remain the author of their own thinking
These are not AI skills. They are human skills. The tools just make them more visible.
A library model for language tools
I keep coming back to a simple idea: the library model.
Publicly funded access. Trained human facilitators. Clear ethical boundaries. Privacy protections. An educational mission instead of a growth mandate.
We didn’t solve literacy by selling books to those who could afford them. We solved it by deciding that knowledge should live somewhere everyone could enter.
Language models are not books. But the moment feels similar.
Equity includes the right to opt out
One last thing matters: equitable access also means respecting refusal.
Some people won’t want to use these tools. Others will want strict limits, cultural boundaries, or human-only spaces. Equity isn’t universal adoption, it’s universal choice.
The goal isn’t a future where everyone uses AI. It’s a future where no one is excluded because they can’t.
That feels like a library problem. And libraries, historically, are one of the better things we’ve built.
None of this erases the real costs of building and maintaining these systems—energy, labor, infrastructure, and environmental impact. Treating something as a public good doesn’t make it free; it makes the costs visible, shared, and accountable.
As someone whose access to work, communication, and participation has narrowed over time, I’m sensitive to the moment when a support becomes a privilege. And I wonder if we are at a precipice.
Leave a Reply