gkcnlr

> "LLMs expose the literacy gap between great and mediocre coders. If you can't see how bad the generated code is, LLMs will only make you not see it faster."

I wholeheartedly believe that if the general trend in SWE goes without hiring new grads/self-taught people (who can still use AI-assistance in their work), we're gonna be seeing huge outsourcing to AIaaS companies and generally become tremendously dependent on them when existing pool of seasoned programmers goes extinct and the work doesn't intrigue training new people.

Although there is cheap, open source models running locally or available at scale on reasonable prices without compromising programmer integrity.

I think the reason why web technologies that are essentially being ported to desktop is not only because they provide cross-platform compatibility (still lacking a standardization on Linux), but they tremendously helped the hustle culture built around the motto of "ship it fast" by providing the fastest iterations and benefiting from already existing cheap pool of web developers, especially via outsourcing/offshoring.

Nowadays, AI (LLMs, genAI, call it what you will) has become so cheap to handle more than one codebase, they no longer feel the need for web only approaches.

The sole reason LLMs has gotten this good is because their aggressive expansion has been accompanied with literally this. The amount of uncredited information these models have been fed must be astonishing, also terrifying.

> [...] as he warned it could trigger a large-scale incident this year or next, such as a temporary blackout of financial or power systems.

I don't buy this. There isn't some kind of voodoo "rogue AI(s)" out there which could randomly decide to take over their vendor stack or brute force their way coming into possession of the most critical infrastructures on earth. These kind of predictions could only be realized if a tiny minority who owns a huge amount of compute and a particular agenda decides to do this kind of thing, which also will require insane amount of data center capability to cause, especially some kind of financial "collapse".

I've always been a fan of SLMs which fulfill a particular role in a constrained environment without ever requiring you to know that you are using an actual AI. IMHO, SLMs accompanied with local, decentralized implementation could really achieve something going forwards instead of a big one model sold by AIaaS providers.

Notice the pattern: It begins after ~70s oil shock and coincides with the rise of neo-liberalism, upon its promise which millions were raised to believe in: "Skilled workers will always be desirable, grill yourself forever and college is the only way that you can be recognized and labeled fit as one."

But many failed to see that this premise would result in one way, which is actually what's happening right now: Growing outsourcing (especially observable in SWE due to its nature) purely driven by market competitiveness (inevitable dynamics of neo-liberalism) and hype surrounding it (leaving no room for other players in the game to choose any other way) which can no longer be sustained through traditionally accepted way of hiring new people who can master new inventions. Although, I highly doubt this is also sustainable.

Eventually, it was well expected for people to stop believing in a system which offers very little foreseeable success of accomplishment and thus, doing manual/repetitive labor is far more preferable for people.

> Looked what happened when crane was invented and not only more houses are built faster but also high rise building suddenly become feasible with more human being employed in the construction industry than ever.

New inventions eventually benefiting the society has always been accompanied by new jobs being created on top of those inventions, a culture of trust built upon mastering those inventions and a market being formulated around demand and supply cycle for those inventions. It is different now[1], because we neither see incentives being created for humans to operate/manage AI-led jobs (because utilizing such a thing is far more unfavorable than outsourcing due to economic incentives, duh) nor a foreseeable market stability for ever such thing to occur.

[1]: https://www.amazon.com/Technofeudalism-Killed-Capitalism-Yan...

Of course, when we talk about utilizing AI as an efficient tooling, it is well expected to see the very outcomes you desire. But the point here is the disconnect between the perceived productivity gains and their corresponding measuring metrics. Remember the Goodhart's Law[1]? It becomes fairly unreliable when you try to measure productivity by a set of standards instead of actually producing productivity. When it comes to future about this issue, I'd hope to see the progress actually happening in open source, decentralized AI instead of the current way of using it.

[1]: https://en.wikipedia.org/wiki/Goodhart%27s_law

How are all these related news regarding silicon shortages surprising at all? It was well predicted that while 2023 chip turmoil[1] was happening, big AI aimed to secure almost all of the demand[2] intended for consumer markets regarding RAM chips so as to win the AI race going forward from that date. They cannibalized almost the entirety of the supply chain, so that they could sell you the incredibly subsidized AIaaS[3] with still considerably higher inference and astronomically higher training process without as of late 2026, failing to achieve any meaningful productivity at all[4]. The serfs now will be happy to own their decade old DDR4 architecture.

[1]: https://www.electronicdesign.com/technologies/embedded/artic...

[2]: https://www.reuters.com/technology/nvidia-supplier-sk-hynix-...

[3]: https://arxiv.org/abs/2606.24616

[4]: https://www.apollo.com/insights-news/insights/daily-spark/No...

> i f*king love LLMs for generating code

> i f*king detest reading LLM-generated material

Thats quite a dichotomy right there.

People just like to spill out their justifications on how they interact with ai and generally try to be agreeable in their comfort zones until they get bamboozled by actual market demands. No competitive structure is willing to pay more (to you) that could be replaced by lower cost means to maximize profits. This is basic economy 101. This article seems a huge cope to me IMHO regarding the advice it gives. Programming is mostly solved, although in a very anti-humane way IMHO. Hobbyist means could justify it in the remaining time...

I just pray open weight wins at the end and people actually utilize their ai and own their ai.

Two more weeks, then it'll happen.

Learning process must be accompanied with enough space for an individual to realistically challenge himself to actually acquire that knowledge with foreseeable success of doing it. A person must believe s/he can really learn something. But in this status quo of AI hypeism (believing that knowing refined set of know-hows would be enough) people gradually begin losing that optimistic "a priori" belief to learn things.

But another side effect of this process is if people stop believing in the accrual of the knowledge which will lead them to nowhere (use cases for those information becoming irrelevant by the time you start using it), justifying this mentally exhausting practice becomes really hard. And I don't mean this on learning static information but also on how to define and build a meta-cognitive framework to systemically learn things.

I don't quite get the "lifelong learning" approach, since lifelong learning must usually be accompanied with organic evolution of the information space. Employers wont pay you because you are a lifelong learner, they'll pay you to actually fix a problem you provide a solution for. And that solution isn't guaranteed by the knowledge you possess and doesn't usually qualify the marginal cost of being a lifelong learner.

One could brain storm on these issues by critiquing the premise of this book: https://www.amazon.com/100-Year-Life-Living-Working-Longevit...

It seems like everybody is focused on "LLM"s, a.k.a Large Language Models. One interesting addition to that is fine-tuned- small parameter, distilled, context-dependent small language models that:

1- Do a particular task with great capability (due to its constrained, limited scope) 2- Do it in such a way, it integrates gracefully in your workflow without ever requiring you to know you are using an LM.

There is a difference between outsourcing your workflow to AI and actually utilizing it.

Check this: https://www.distillabs.ai/blog/we-benchmarked-12-small-langu...

Search for this: "Revenue, classified by significant product and service offerings"

By looking at their 2025 shareholder report (Look for the part below "NOTE 18"), Windows is only at the 5th place in terms of revenue source, even below the LinkedIn:

https://www.microsoft.com/investor/reports/ar25/index.html#

I can only think that they do not even care about Windows anymore, let alone Notepad...

As long as AI (genAI, LLMs, whatever you call it describe the current tech) is perceived not as a "bicycle of the mind" and a tool to utilize 'your' skills to a next phase but as a commodity to be exploited by giant corporations whose existence is based on maximizing profits regardless of virtue or dignity (a basic set of ethics to, for example, not to burn books after you scan it feed your LLM like Anthropic), it is really hard to justify the current state of AI.

Once you understand the sole winner in this hype is the one who'll be brutally scraping every bit of data, whether it's real-time or static and then refining it to give it back to you without your involvement in the process (a.k.a, learning) you'll come to understand that the current AI by nature is hugely unfavorable to mental progression...

Current economic supply/demand balance as well as philosophical principles are so utterly deranged, people lean toward what could maximize their profit. If this is achieved through conformism & doing what you're accustomed to, no one will try to truly innovate and prefer staying in their comfort zone.