After we speak about AI, we are inclined to deal with outcomes: what it might do, the place it’s going, the way it’s outperforming people in activity after activity. However far much less consideration goes to what feeds these programs — and what which means for the folks behind the info.
As a result of AI doesn’t simply study from details. It learns from us. From our language, our clicks, our routines, our creations. From posts scraped with out consent. From discussion board threads and pictures and even medical datasets many by no means knew had been getting used.
In 2024, The Atlantic revealed how a lot of its archive — going again many years — was used with out authorization to coach industrial AI fashions. Reddit, StackOverflow, X (previously Twitter), and numerous boards adopted go well with. In Might 2024, a category motion lawsuit was filed in opposition to OpenAI for allegedly coaching ChatGPT on non-public information, together with emails and chats, with out customers’ information or consent.
These are pressing copyright and digital consent points. A few information financial system more and more constructed not on participation, however extraction.
The Phantasm of “Decide-In”
We stay in a world the place most individuals by no means actively agreed to their information coaching massive language fashions. However now, that information is encoded, weighted, and regurgitated by AI instruments that form search engines like google, hiring choices, advert concentrating on, and even inventive industries.
It’s a quiet type of dispossession: the normalization of being mined, modeled, and mimicked by programs you don’t management and sure by no means will.
What We Threat Dropping
If AI turns into the dominant interface of the web — mediating what we see, how we work, and the way we talk — then who trains it, and the way, turns into a matter of energy.
When information is centralized, historical past turns into editable and when programs bear in mind every little thing, your freedom on-line begins to sound the hazard alarm.
That’s why AI literacy goes past learn how to use instruments like ChatGPT or Midjourney and offers with the very boundaries that we, the folks, the customers, are conscious of and vigilent sufficient to talk for.
Listed below are some widespread sense questions all of us ought to be asking:
Who owns the info AI learns from?Who decides which info is emphasised or erased?What rights do creators, educators, and residents have over their enter?Can AI be educated on moral constraints — and who defines these ethics?
And most critically: what infrastructures are we constructing to help clear, decentralized, and self-determined fashions?
Our Place at SourceLess Labs Basis
We imagine AI ought to serve human dignity, not override it. And that begins with infrastructure — the place id, information, and computation are usually not trapped in walled gardens.
For this reason SourceLess builds:
Non-public computation frameworks the place AI brokers function transparently and serve their customers, not simply the businesses behind them.Verifiable digital identities by STR.Domains, the place the person owns their credentials — moveable, encrypted, and never issued by a third-party app.Decentralized studying and collaboration areas — so creators and educators aren’t compelled to commerce privateness for entry.
We imagine human literacy on this new period should embrace infrastructural consciousness not simply learn how to use instruments, however how they’re made, maintained, and monetized.
As a result of in the long run, the programs we practice will replicate not simply our inputs however our intentions.