To think or not to think

Year 2030, Ai has become so expesive that all we can afford are chinese models. There are plans that offer discounts if you give up the data in the inference contexts, it’s not clear if this is legal, people fill in a huge legal agreement that makes it hard to contest.

In general the jurisdictions around the world had to loosen up intellectual right ownership. There are many class actions in which major providers settle with big money, in the end everything that is reachable on the internet has been declared public, and usable to train models.
In border line scenarios where it’s not clear how a specific set of training data entered the model, if you can prove the ownership of it, a sort of bounty is awarded. But it’s a difficult process that few are capable to persue.

Having a frontier model for coding is not a priority anymore, develepors with decent computers can easily use local models. Companies selling the hardware diversified investments in open source/weight models and are very happy to supply their services.

Mental strength

What organizations had envisioned in 2026, that lots of workers would have been replaced by AI didn’t deliver. But humanity in general got dumber, remember when people started to use map apps and lost the capability to orient themselves on the road?
The process to make people stupid changed level with LLMs, this was a process that had already started with the doom scrolling phenomenom in socials, and short contents polluting focus space. With afterthought this was predictable, supply a tool that suggests the next possible token, and human stop exercising that capability to weigh options and make calls indepentdently. In working environments the knowledge workers now have to pay the extra toll for a tool that suggests solutions and the human in the loop is a passthrough orchestrator, too slow to work without it, but since the tool is didn’t develop the awareness his human partner there binded together for eternity.
Speed is the name of the game, but the most profilic workers are the ones that have retained the capability to slow down.

Code Refactorings

In apps under development it is common to rewrite the codebase, usually the first step is to rewrite it in Rust. Then you can switch frameworks, writing style, usually in the end everybody usually lands on isomorphic Typescript. Test harnessing is full of slop, that’s generally tollerated, and you can rewrite that too. The general pattern that works is to divide in components with clear boundaries, nevertheless incidents are common and burning out is a cyclical phase.

Acceptance

Society is progressively loosing it’s human centricity. Nobody talks about 40 hours working week anymore, the new norm is to work at least 80 hours, but once a week people go out to party and most of the time someone get gun wounded. Having lost the capability to think people are also less incline to establish deep relationships, even if there is perception that something’s wrong only few are able to process it, most just react with uncontrolled animal behavior and then find peace in drugs. Overall there is no revolution, “we are where we are” is accepted.

World order

There was once a narrative of a group of people dictating a world agenda that should have driven humanity towards something. Don’t know if the coorporate purpose changed, but it feels that the only thing that matters now is company efficiency, big mega corps led by founders persueing different visions, but with one commonality: the org needs to make a lot of money to fuel their dreams.

The message

Remember you must die. It’s funny how often i find this statement a relief, don’t get me wrong like most of my contemporaries fear of death is still a real thing. Survival instinct will never go away as long as we rember the feeling of a brath of fresh air. Just keep on thinking, there might be a time where “you” might be just an uploaded version of yourself, your thoughts will always be the most precious assets.